Chapter 10: Automated versus Autonomous Systems
Walk into a large shop and the glass door slides open before you touch it. A sensor above the door notices that someone is standing close, and the door opens. The same thing happens every single time. It does not matter whether you are carrying bags, walking quickly, or just standing near the entrance and talking to a friend. Come close, the door opens. Move away, the door shuts.
Now look at the floor-cleaning machine moving quietly between the shelves. It is not following a painted line. It turns away from a trolley someone has left in the aisle. It goes around a wet patch. When its battery runs low, it finds its own way back to its charging point. Nobody is steering it.
Both machines work without a person pressing a button. But they are not the same kind of machine at all. One simply follows a fixed rule. The other has to work out what to do as it goes, in a place that keeps changing.
Diagram
Fig. 10.0-1 – a single shop scene split into two halves. Left half: an automatic sliding glass door with a sensor box above it and a shaded detection zone on the floor where a shopper is standing, labelled “same input, same action”. Right half: a small floor-cleaning robot inside the shop steering around a trolley and a wet patch, with a dotted path showing it curving past both obstacles, labelled “works out its own path”
This chapter will answer all five of these questions.
10.1 Automated Systems
Think of a job that has to be done the same way, hundreds of times a day. Filling exactly 200 millilitres of oil into every bottle that passes on a belt. Stamping the same shape out of one metal sheet after another. Opening a gate every time a bus arrives at a depot.
A person can do all of these. But a person gets tired, and a tired person makes mistakes. The four-hundredth bottle does not get the same care as the first one. The work is dull, it is repetitive, and none of it needs any real thinking — the rule is the same every time.
This is exactly the kind of work we hand over to a machine. An automated system is a set of technologies or processes that operates with minimal, or sometimes no, human intervention. It follows rules that were fixed in advance, and it carries out tasks based on programmed instructions or inputs.
An input is simply something the machine receives from outside — a button being pressed, a sensor noticing an object, a timer reaching zero. The machine reads the input, checks it against its rules, and acts.
The pattern behind every automated system
Every automated system has the same shape underneath, whatever it is made of. Something comes in, a fixed rule is applied to it, and an action follows. See Fig. 10.1-1.
Diagram
Fig. 10.1-1 – a simple horizontal block diagram with three boxes joined by arrows: “INPUT (button pressed, sensor triggered, timer ends)” → “FIXED RULE set by a person in advance” → “ACTION (motor turns, valve opens, light changes)”. Below the whole chain, a caption strip reading “The same input always produces the same action”
The important part of that picture is the middle box. The rule inside it was written by a person, long before the machine was switched on. The machine does not add to that rule or change it. So the same input will always produce the same action, today and next month.
Notice also what the word “automated” does not mean. It does not mean the machine looks after itself completely. A person designed it, wired it, set the values it works with, switched it on, and will come back to check it. An automated system typically requires some level of human insight and intervention. What it removes is the need for a person to stand there and do each repetition by hand.
Because the rules are fixed, an automated system needs a controlled and predictable environment — a factory floor, a packing line, a lift shaft — where conditions stay roughly the same and surprises are rare.
Everyday examples
- Assembly line robots in a factory, doing the same welding or painting movement on every car body that arrives.
- Automatic doors at a shop or an airport, opening whenever a sensor detects someone in front of them.
- Industrial machines that run through a preset sequence of operations — cut, press, cool, eject — over and over.
- A lift that takes you to the floor whose button you pressed, using the same rules for every passenger.
What an automated system cannot do
An automated system lacks the ability to make independent decisions. It follows its set of instructions without taking changing conditions into account. If something happens that the rules never mentioned, the machine does not stop and think about it — it simply carries on with the rule it has.
That is fine on a factory floor, where the situation barely changes. It becomes a real limitation the moment the surroundings become unpredictable.
There is a second limitation, and it matters just as much. If you want the machine to behave differently — fill 500 millilitres instead of 200 — you cannot explain that to it. Someone has to reprogram it or adjust it by hand. The machine does not improve with practice, and it does not learn anything from the thousands of bottles it has already filled.
So automation is a very good answer to repetitive work in a steady environment, and a poor answer to work in a place that keeps changing. That second kind of work needs a different kind of system, which is where we go next.
10.2 Autonomous Systems
Now take the same idea out of the factory and put it somewhere untidy. A hospital corridor, for instance, where a small robot has to carry a tray of medicines from the pharmacy on the ground floor to a ward on the second floor.
No two trips down that corridor are the same. A trolley is parked in a different spot each time. A group of visitors is standing near the lift. A patient is walking slowly with a stick. Someone has left a bucket out after mopping.
A fixed sequence of instructions is useless here. Nobody can write down, in advance, a rule for every obstacle that might appear in a corridor. What is needed is a machine that keeps sensing what is actually in front of it and works out its next move from that.
That is an autonomous system — a system capable of performing tasks or making decisions without human intervention. It operates independently and can adapt to changing environments. It handles complex, changing situations by assessing conditions in real time and choosing its actions from programmed logic, artificial intelligence (AI) — the branch of computing that builds systems able to do tasks we normally think of as needing human intelligence — or machine learning (ML) algorithms, which are methods that build their rules from data instead of being given the rules directly.
A word of caution about the phrase “making decisions”. The robot is not thinking about the bucket. It is comparing sensor readings against the logic inside it and computing which of its possible actions best fits what it measured. We call that a decision because it is a genuine choice between options — but the choosing is arithmetic, not thought.
The loop an autonomous system runs
An automated system runs in a straight line: input, rule, action. An autonomous system runs in a loop, and never stops going round it while it is working. See Fig. 10.2-1.
Diagram
Fig. 10.2-1 – four boxes arranged in a circle with arrows going clockwise: “SENSE the surroundings”, “ASSESS the conditions right now”, “CHOOSE an action”, “ACT”. A return arrow from ACT back to SENSE closes the loop. A label beside the loop reads “runs continuously, many times a second”
The loop is the whole difference. Because the machine senses again after every action, a change in the corridor changes what it does next. The bucket that was not there yesterday is simply part of today’s reading.
Everyday examples
- Autonomous vehicles that navigate roads without a human driver.
- Drones that carry out tasks such as delivery without direct control from a pilot.
- Robotic vacuum cleaners that find their own way around a room and clean it.
What autonomy costs you
Autonomy is not a free upgrade. Three things get harder.
The system becomes much more complex to build. It needs sensors, a lot more computing, and far more careful software than a machine that follows one fixed sequence.
Its behaviour also becomes harder to predict. Run the corridor robot twice and it may take two slightly different paths, because the corridor was not identical on the two runs. That is exactly what you wanted — but it makes the machine harder to test, because you can no longer check it by confirming that it always does the same thing.
And its judgement is only as good as its information. Blocked sensors, poor light or a bad map will produce poor choices, and the machine has no way of knowing that its picture of the world is wrong.
10.3 Comparing Automated and Autonomous Systems
Both kinds of system do their work without a person operating them by hand. That is why they are so easily mixed up. The difference is not in how much a person touches the machine — it is in where the machine’s next action comes from.
In an automated system, the next action comes from a rule written earlier by a person. In an autonomous system, the next action is computed from what the machine is sensing right now. Everything else follows from that one difference, as Table 10.1 shows.
Read the table as one idea told five times, not as five separate facts to be memorised. Fixed rules are cheap and easy to check, which is why automated systems stay simple. Working things out live is expensive and hard to check, which is why autonomous systems are complex. Every row comes back to that same trade.
Not always a sharp line
Real machines do not always sit neatly in one column. A modern car can hold a set speed on the highway by following a fixed rule, and at the same moment brake because a camera has picked up something ahead. One part of it is automated; another part is autonomous.
So the more useful question is how much of a machine’s behaviour was fixed in advance, rather than which single label it deserves. Answering that means looking inside, at how the rules reach an answer. That is what the rest of the chapter does.
10.4 Deterministic Decision Making
We have sorted machines by how much they run on their own. Now we open them up and look at how the answer is actually reached inside. Two systems can look equally independent from the outside and still work in completely different ways within.
Start with the simpler of the two. Think of the gate at a paid parking area. A car leaves, and the machine must produce a fee. There is nothing to weigh up here — the parking rate was fixed by the owner, the entry time was recorded, and the fee follows from a short calculation. Two cars that stayed for the same length of time must be charged the same amount, or customers will rightly complain.
This is deterministic decision making: a process in which decisions are made from predefined rules, without any randomness or uncertainty involved. Give the system the same starting conditions and the same inputs, and it will always produce the same outcome.
The method a machine follows to get from its inputs to its answer is called an algorithm — a finite list of clear steps that always finishes and always gives an answer. In a deterministic system, the algorithm is written by a person, stored in the machine, and does not change unless a person changes it.
The same rule drawn as a flowchart is shown in Fig. 10.4-1. Follow any starting time through it and you will always come out at one exit.
Diagram
Fig. 10.4-1 – a simple flowchart. Start box “Car reaches exit gate” → process box “time parked = exit time − entry time” → process box “fee = 20” → decision diamond “any time left beyond the first hour?” with a “No” arrow going straight to the end and a “Yes” arrow going to a process box “add 10 for each further hour or part hour” which then joins the end box “display fee”. Mark the single Start and single End clearly
That last sentence is the whole point of a deterministic system. Repeatability is not a side effect here — it is the reason the approach was chosen.
Deterministic decision making is straightforward and transparent, which makes it valuable in controlled settings where the rules and conditions are well understood. It is the right choice wherever consistency, predictability and reliability matter more than cleverness.
But it has a real weakness. A deterministic system assumes that all the information it needs is available and correct — and in the real world, that is often not true. Faced with a situation its rules never anticipated, it cannot adapt. It applies the nearest rule it has and produces an answer that may be badly wrong.
10.4.1 Key Characteristics of Deterministic Decision Making
Four features describe every deterministic system, and each one is really the same property seen from a different side.
- Predictability: The outcome is always the same when the input conditions are identical. Nothing in the process varies from one run to the next.
- Rule based: Decisions come from a specific set of rules, formulas or algorithms, which do not change unless someone deliberately modifies them.
- No uncertainty: The process assumes that everything relevant is known and that the surroundings are stable, with no unforeseen event affecting the outcome.
- Transparency: The reasoning behind any decision can be followed and understood, because it is a logical progression through the given rules.
Transparency is worth pausing on. If a driver disputes the Rs 50, the operator can walk through the flowchart line by line and show exactly where the amount came from. Very few systems can be explained that completely, and in matters like billing, safety and law, being able to explain an answer is often as important as the answer itself.
10.4.2 Where Deterministic Decision Making Is Used
The approach turns up wherever the rules are known in advance and the same situation is expected to give the same result.
- Manufacturing: Automated systems follow precise steps so that every item leaving the line is identical to the last one.
- Finance: Algorithms that carry out transactions when specific, stated criteria are met, without deviating from them.
- Software development: Ordinary programs, where a given input leads to one well defined output.
- Everyday appliances: A thermostat that switches a heater on or off at a set temperature threshold, and similar controls in a refrigerator or an air conditioner.
- Expert systems: Programs that reach a conclusion by applying a series of predefined rules to the data they are given.
Calculations are the purest case of all. Working out the area of a rectangle from its length and its width is deterministic by nature: the same two measurements can only ever produce one area.
10.5 Probabilistic Decision Making
The parking gate had it easy. Everything it needed was already known — the entry time, the exit time, the rate.
Now try a different question. A school has to decide, the evening before, whether to set up the stage outdoors for the annual day or move it into the hall. The answer depends on whether it rains tomorrow, and nobody knows whether it will rain tomorrow. No rule can be written that turns today’s sky into a certain answer about tomorrow, because the information simply does not exist yet.
There is still something useful to work with, though. We know what happened on many past days that began like this one. That does not give certainty, but it gives a likelihood — and a likelihood is enough to make a sensible choice.
This is probabilistic decision making: a process where decisions are made from the likelihood, or probability, of different outcomes. Instead of following a fixed set of rules, it takes the uncertainty and variability in the environment or the data into account, and uses statistical models or probability distributions to assess the possible outcomes.
Probability is just a number that says how likely something is. It runs from 0, meaning it never happens, to 1, meaning it always happens, and is often written as a percentage instead — 0.7 and 70 per cent mean the same thing.
Where the number comes from
A forecast is not a guess. It is counted from records. Table 10.2 shows a small made-up record of ten past days whose morning conditions matched today’s.
The probability is worked out from those counts:
\[P(\text{rain}) = \frac{\text{number of similar days on which it rained}}{\text{total number of similar days}}\]
Here \(P(\text{rain})\) means the probability of rain, the top of the fraction counts the “Yes” rows, and the bottom counts all the rows.
Read that answer carefully. It does not say it will rain. Seven days out of ten like this one ended in rain, and three did not — so a dry evening is still entirely possible. The decision was not made because the outcome was known. It was made because one outcome was clearly more likely than the other, and getting the stage wet costs more than moving it indoors for nothing.
What this approach costs
Probabilistic decision making needs more sophisticated mathematical models and far more computing than a fixed rule does. Probabilities are also easy to misread — people often hear “70 per cent chance” as “it will happen” — and that misreading can be dangerous in high-risk situations.
Above all, it leans entirely on the model and the data underneath it. If the record of past days is small, out of date, or collected carelessly, the probability that comes out of it will be misleading, and the decision built on it will be poor.
10.5.1 Key Characteristics of Probabilistic Decision Making
- Incorporation of uncertainty: The system accepts that not everything is known or certain, so decisions are made on probabilities rather than on certainties.
- Risk assessment: It weighs the risks and the benefits of the different options, using the likelihood of each outcome.
- Flexibility: It can adapt to changing information by updating its probabilities and reconsidering a decision as new data arrives.
- Use of probabilistic models: It usually relies on statistical methods, Bayesian inference or machine learning algorithms to estimate how likely each outcome is.
Flexibility is the characteristic that most clearly separates this approach from the last one. A deterministic rule stays exactly as it was written. A probabilistic model shifts as fresh data comes in, which makes it far more responsive to a changing environment.
10.5.2 Where Probabilistic Decision Making Is Used
The approach suits any setting where uncertainty is built in and not all the information is available.
- Weather forecasting: Predicting the chance of rain from past records and current atmospheric conditions, so that people can decide whether to carry an umbrella.
- Financial investment: Weighing the possible returns and risks of different options before putting money into any of them.
- Robotics and AI: Self-driving cars and similar systems making real-time choices from sensor data that is never perfectly accurate.
- Medical diagnosis: Doctors reasoning about how likely various conditions are, given a patient’s symptoms and history.
- Business strategy: Companies judging market trends and customer behaviour before launching a product or setting a price.
- Engineering: Reliability engineering, which predicts how likely a system is to fail and plans maintenance around that.
- Artificial intelligence: Machine learning and natural language processing, which use probabilistic models to make their predictions.
10.6 The Four Kinds of Systems
We have now asked two separate questions about a machine, and it is worth noticing that they really are separate.
The first question was: does it need a person, or does it run on its own? That gave us automated and autonomous systems. The second was: does it reach its answer by a fixed rule, or by weighing likelihoods? That gave us deterministic and probabilistic decision making.
Neither question decides the other. A machine that runs entirely on its own may still follow rules that never change. A machine that needs a person to set it up may still work in probabilities inside. Cross the two questions and you get four kinds of system, as shown in Fig. 10.6-1.
Diagram
Fig. 10.6-1 – a two-by-two grid. The columns are labelled “AUTOMATED (needs a person to set up, start or adjust)” and “AUTONOMOUS (runs and chooses actions on its own)”. The rows are labelled “DETERMINISTIC (fixed rules)” and “PROBABILISTIC (weighs likelihoods)”. The four cells read, clockwise from top left: “Deterministic automated — traffic light on a fixed cycle”, “Deterministic autonomous — warehouse vehicle on a fixed route”, “Probabilistic autonomous — self-driving car”, “Probabilistic automated — spam filter”. Draw an arrow along the bottom of the grid labelled “predictability decreases” pointing right, and an arrow up the left side labelled “handles uncertainty better” pointing down
Each of the four is described below. As you read them, keep asking the two questions — who starts it, and how does it choose — because those two answers place any machine you meet.
10.6.1 Deterministic Automated Systems
A deterministic automated system is an automated system that works from a set of predefined rules or algorithms and produces predictable, consistent outcomes. For a given set of inputs or conditions, it always generates the same output, or follows the same sequence of actions, with no variation and no randomness.
The defining feature is that these systems require an external trigger to operate. Something outside must set them going — a coin dropped in, a timer running out, a button pressed. Left alone with nothing happening around them, they do nothing.
Examples:
- Traffic light systems: Traditional traffic lights run on a fixed schedule, changing the lights at predetermined intervals no matter what the traffic is actually doing at that moment.
- Vending machines: A machine that releases one particular item when the correct amount is inserted and that button is pressed, in exactly the same way for every customer.
These systems cannot adjust to a new or unexpected situation, which limits where they can be used. They also do not improve with time. A traffic light that has controlled the same junction for ten years handles this evening’s rush exactly as it handled the first one, because nothing about running for ten years changes its schedule.
10.6.2 Deterministic Autonomous Systems
A deterministic autonomous system is an autonomous system that works according to fixed, predefined rules, and whose behaviour stays predictable and consistent for the same set of inputs or conditions. Unlike a deterministic automated system, it does not wait for an external trigger. It carries out its task and takes its actions on its own — but always in a predictable way.
The rules are programmed in, and they do not change unless a human operator alters them. Under the same conditions, the system performs the same task in the same way every time, which is exactly what makes it reliable. What it cannot do is learn. It does not change its behaviour with experience, and it cannot cope with an input nobody anticipated when it was designed.
Examples:
- Automated Guided Vehicles (AGVs): Vehicles in a warehouse that carry goods from one point to another along fixed routes, without a human driver, but always over the same paths.
- Drones with predefined paths: A drone programmed to fly a set route for aerial photography or surveying, operating independently but covering the same path on every flight.
10.6.3 Probabilistic Automated Systems
A probabilistic automated system is an automated system that makes decisions or performs tasks using probabilistic models or statistical methods, instead of fixed, deterministic rules. It generally needs an initial setup and occasional adjustment or supervision, but once configured it runs automatically.
These systems are built to handle uncertainty in their inputs, their environment or their processes. They use probabilities to decide, which lets them respond flexibly to conditions that vary. This has an odd-looking consequence: given the same inputs, such a system may produce different outputs, depending on the probabilities involved. That is not a fault. It lets the system explore several possible actions rather than being locked into one predetermined path.
Some probabilistic automated systems go further and adjust their models from past data, so their decisions improve as more information is gathered.
Examples:
- Spam filters: An email system that estimates how likely each arriving message is to be unwanted, from patterns it has learnt across many past messages, and sorts it accordingly. Nobody sorts each message by hand, but a person set the filter up and can retrain it.
- Robust control systems: Automated controls in industrial processes that use probabilistic methods to cope with uncertainty in their measurements or in the environment, so performance stays good under varying conditions.
- Data-driven decision making: Systems in finance, healthcare and marketing that analyse data and make predictions under uncertainty.
These systems are harder to design and build than deterministic ones. They need sophisticated algorithms and real computing power, and because their outcomes vary, their behaviour is harder to predict and to check.
10.6.4 Probabilistic Autonomous Systems
A probabilistic autonomous system is an autonomous system that makes decisions and performs actions using probabilistic models or algorithms. This lets it handle uncertainty, adapt to changing environments, and make predictions from incomplete or uncertain information.
It is the most capable of the four and the most demanding to build. It runs with minimal or no human intervention, makes its choices in real time from its probabilistic analysis, learns from its environment, and changes its behaviour accordingly.
Examples:
- Self-driving cars: Vehicles that make real-time choices about navigation, obstacle avoidance and traffic by estimating how other vehicles and pedestrians are likely to move next, and adjusting their own actions to those estimates.
- Autonomous drones: Drones used for delivery, surveillance or mapping, which navigate changing surroundings, avoid obstacles and adjust to varying weather.
Everything about such a system rests on the quality of its data. Poor or biased data leads directly to poor or wrong decisions, and the system cannot tell that this is happening — it will act on a mistaken estimate with exactly the same confidence as a good one.
10.6.5 Choosing Between the Four
More capable is not the same as better. Each of the four is the right answer to a different problem, and picking the wrong one is expensive in both directions.
Ask what the surroundings are like. If conditions stay steady and predictable, fixed rules will serve, and paying for sensing and probability buys nothing. If conditions change constantly and cannot be listed in advance, fixed rules will fail sooner or later.
Ask how much the system’s answers must be explainable. Where a decision has to be justified to a customer, an inspector or a court, a deterministic system can be walked through step by step. A probabilistic one can only report how likely it thought each outcome was.
Ask what a wrong answer costs. Where the same situation must always produce the same result — billing, safety interlocks, examination marks — variation is a defect, not flexibility. Where being roughly right under uncertainty beats being exactly right only when everything is known, probability wins.
And ask what you can afford to build and test. A deterministic automated system can be checked by confirming that it always does the same thing. Testing a probabilistic autonomous one is a far larger job, because it may never do quite the same thing twice.
10.7 Summary
- An automated system carries out tasks by following predefined rules, using programmed instructions or inputs, with minimal or sometimes no human intervention while it runs.
- An automated system still needs some human involvement — someone designs its rules, sets it up, starts it and adjusts it. It suits a controlled, predictable environment and repetitive work.
- An automated system cannot make independent decisions, cannot adapt to conditions its rules never anticipated, and does not improve with experience.
- An autonomous system performs tasks and chooses its actions without human intervention. It senses its surroundings in real time, assesses conditions and adapts to a changing environment.
- An autonomous system runs a continuous loop — sense, assess, choose, act — using programmed logic, artificial intelligence or machine learning algorithms.
- The two differ in human intervention, operating environment, flexibility, decision making and complexity. Fixed rules make a system simple and easy to check; working things out live makes it capable but complex and harder to test.
- Deterministic decision making follows fixed, predefined rules, with no randomness or uncertainty. The same inputs always give the same outcome.
- Its key characteristics are predictability, being rule based, no uncertainty, and transparency. It is used in manufacturing, finance, software development, everyday controls such as a thermostat, and expert systems.
- Deterministic decision making assumes all the needed information is available and correct, so it copes badly with new or uncertain situations.
- Probabilistic decision making weighs the likelihood of different outcomes, using data and statistical models, instead of one fixed rule.
- Its key characteristics are the incorporation of uncertainty, risk assessment, flexibility, and the use of probabilistic models. It is used in weather forecasting, financial investment, robotics and AI, medical diagnosis, business strategy, engineering and artificial intelligence.
- Probabilistic decision making needs more computing and better data. Poor or biased data leads to poor decisions, and probabilities are easily misread as certainties.
- The two questions — automated or autonomous, deterministic or probabilistic — are independent, so they give four kinds of system.
- A deterministic automated system follows fixed rules and needs an external trigger. Traditional traffic lights and vending machines are examples.
- A deterministic autonomous system acts on its own but always in the same predictable way, because its rules never change. Automated Guided Vehicles and drones on predefined paths are examples.
- A probabilistic automated system uses statistical methods once it has been set up, so the same inputs may produce different outputs. Spam filters and robust control systems are examples.
- A probabilistic autonomous system acts on its own and reasons about uncertainty, adapting to changing conditions. Self-driving cars and autonomous drones are examples.
- More capable is not the same as better. The right choice depends on how steady the environment is, how far decisions must be explainable, what a wrong answer costs, and how much testing the system can be given.
10.8 Key Terms
- Algorithm: a finite list of clear steps that always finishes and always produces an answer.
- Artificial intelligence (AI): the branch of computing that builds systems able to do tasks we normally think of as needing human intelligence.
- Automated Guided Vehicle (AGV): a driverless vehicle that carries goods along fixed routes, usually inside a warehouse or factory.
- Automated system: a system that carries out tasks by following predefined rules, using programmed instructions or inputs, with little or no human involvement while it runs.
- Autonomous system: a system that carries out tasks and chooses its actions without human intervention, by sensing its surroundings in real time and adapting to conditions that were not fixed in advance.
- Deterministic automated system: an automated system that follows predefined rules, needs an external trigger, and always produces the same output for the same inputs.
- Deterministic autonomous system: an autonomous system that acts without human intervention but follows fixed rules, so its behaviour stays predictable and consistent.
- Deterministic decision making: decision making that follows fixed, predefined rules, so that the same inputs always lead to the same result.
- Machine learning (ML): a set of methods that build their rules from data instead of being given the rules directly.
- Probabilistic automated system: an automated system that decides using probabilistic models or statistical methods once it has been set up, so the same inputs may produce different outputs.
- Probabilistic autonomous system: an autonomous system that decides and acts using probabilistic models, allowing it to handle uncertainty and adapt to changing environments.
- Probabilistic decision making: decision making that weighs the likelihood of different possible outcomes, using data and statistical models, instead of following one fixed rule.
- Probability: a number showing how likely something is, running from 0 for never to 1 for always, and often written as a percentage.
10.9 Objective Questions
- Fill in the blanks.
- A system that carries out tasks by following predefined rules, with little or no human involvement while it runs, is called an ________ system.
- Decision making in which the same inputs always lead to the same result is called ________ decision making.
- A probability of ________ means that an outcome never happens.
- Which of these is a feature of an autonomous system?
- It repeats a fixed sequence of steps stored in advance
- It senses its surroundings and adapts to conditions that change
- It waits for a person to press a button before every action
- It works only inside a controlled and predictable environment
- State whether each statement is true or false.
- An automated system improves its performance the longer it runs.
- A deterministic autonomous system can act without an external trigger.
- A probabilistic system may give different outputs for the same input.
- A vending machine releases a packet of biscuits when the correct amount is inserted. It is an example of a:
- deterministic automated system
- deterministic autonomous system
- probabilistic automated system
- probabilistic autonomous system
- Match each system in Column A with the correct description in Column B.
| Column A | Column B |
|---|---|
| a. Traffic light running on a fixed cycle | i. Probabilistic autonomous system |
| b. Warehouse vehicle following a set route with no driver | ii. Deterministic automated system |
| c. Email spam filter | iii. Probabilistic automated system |
| d. Self-driving car | iv. Deterministic autonomous system |
- Which of these tasks is best suited to probabilistic decision making?
- Working out the area of a rectangle from its length and width
- Calculating a bill from a fixed rate and a meter reading
- Estimating the chance of rain from records of past days
- Opening a gate when a button is pressed
- A parking gate charges Rs 20 for the first hour or part of it, and Rs 10 for every further hour or part of an hour. What fee will it display for a car parked for 1 hour and 5 minutes?
10.10 Concept Questions
Define an autonomous system in your own words, and give one example that is not from your surroundings at school.
State any three key characteristics of deterministic decision making, writing one line about each.
Differentiate between a deterministic automated system and a deterministic autonomous system. Support your answer with one example of each.
A record of 20 past days with the same morning conditions shows that it rained on 12 of them. Work out the probability of rain as a percentage, showing your steps. Then state one thing this number does not tell you.
Give a reason why a probabilistic system is harder to test than a deterministic one.
A survey drone is programmed to fly the same fixed path over a field every week. Since it was programmed, a tall mobile tower has been built along that path. What would happen on the next flight, and why? What kind of system would have handled this better?
10.11 Case-Based Questions
- A farmers’ co-operative waters its vegetable plots using a sprinkler that switches on at 6 a.m. every day and runs for 20 minutes. It does this whether the soil is dry or already wet, so water is often wasted after a rainy night.
The co-operative is considering a new controller. It would read a soil-moisture sensor and also take the day’s rain forecast into account before deciding whether to water, and for how long. The forecast for the coming week is:
Monday 10 per cent, Tuesday 20 per cent, Wednesday 85 per cent, Thursday 15 per cent, Friday 90 per cent.
- Classify the present sprinkler as automated or autonomous, and as deterministic or probabilistic. Give one reason for each part of your answer.
- On which two days of that week would the new controller most likely reduce or skip the watering, and why?
- State one risk the co-operative takes on by moving to the new controller.
- In the storeroom of an online shop, driverless vehicles carry boxes from the shelves to the packing counter. Each vehicle follows a fixed route that was programmed into it. When two vehicles meet in the same aisle, both stop and wait until a worker moves one of them aside.
The owner wants to replace this with a system in which each vehicle estimates where the other vehicles are likely to be, and chooses a route around them.
- Which of the four kinds of system does the present arrangement belong to? Justify your answer.
- Name two kinds of data the new vehicles would need that the present ones do not use.
- The owner expects fewer stoppages after the change. State one new difficulty the change will bring.