Five challenges in designing fully autonomous systems for self-driving cars

Self-driving cars - this concept has been hyped up a lot in the last eight to ten years. Although a lot of research has been done on this issue as early as fifty years ago, research on how to make cars learn by themselves and then be able to drive themselves has been done.

Today, we see a lot of experiments with driverless cars under human supervision, in controlled environments, and in good road and environmental conditions, but there are several challenges in designing fully autonomous cars. The whole concept of "driverless" vehicles on the road has aroused curiosity among all sectors of society. Not many people believe it is possible yet, or that it will be possible anytime soon.

Why shouldn't it be so? There are many moving parameters in the driving function that need to be processed and controlled simultaneously. Even a single failure can be catastrophic. In reality, autonomous driving There is no absolute black or white. There are roughly five different levels that define the level of automation. At the basic level, someone (the driver) controls all functions such as brakes, steering, throttle, power, etc.

01 Introduction to other levels

Level 1: Most functions are still controlled by a human (driver), but certain specific functions (such as steering or acceleration) can be done automatically by the car.

Level 2: In Level 2, at least one steering and acceleration/deceleration driver assistance system that uses information about the driving environment is automated, such as cruise control and lane centering. As a result, the driver begins to disengage from actually operating the vehicle, with his hands off the steering wheel and his feet off the pedals. However, in this situation, the driver should still remain alert and must be ready to take control of the vehicle at any time.

Level 3: Level 3 cars still require a human driver, but in certain traffic or environmental conditions, safety-critical functions can be fully transferred to the car. This means that the driver is still present and will intervene when necessary, but does not need to monitor the situation in the same way as the previous level.

Level 4: Level 4 refers to “full autonomy.” Level 4 vehicles are “designed to perform all safety-critical driving functions and monitor road conditions throughout the journey.” However, this is again limited to the vehicle’s Operational Design Domain (ODD), meaning it does not cover all driving scenarios.

Level 5: This level refers to a fully autonomous system in which the vehicle is expected to perform on par with that of a human driver in every driving scenario, including in extreme environments such as dirt roads that are unlikely to be navigated by driverless vehicles in the near future.

Now comes the real challenge. How to design an autonomous/driverless vehicle system that can handle vehicle performance like a human in all possible conditions. Autonomous vehicles are usually a combination of sensors and actuators , complex algorithms and powerful processors that execute software . There are hundreds of such sensors and actuators, located in various parts of the vehicle, driven by a highly complex system.

02 The perception system can be divided into three different parts

Navigation and guidance: Systems for determining where you are, where you want to go, and how to get there. Compass, sextant, LORAN radio Instruments and techniques such as positioning and dead reckoning have been used, with varying degrees of accuracy, consistency, and availability.

Driving and safety: guiding the vehicle, ensuring it drives properly in all situations and obeys the rules of the road. An autonomous car must be able to see and interpret what’s ahead when driving forward (and, of course, behind when reversing). It’s also necessary to see what’s to either side; in other words, it needs a 360-degree⁰ view. Obviously, an array of cameras could be chosen, with cameras determining the position of lanes and sensing objects or markings on the road.

Performance: Managing the car’s internal systems, a large part of autonomous vehicle design involves mundane issues such as power management. Several unique, application-specific boards and subsystems are added to traditional vehicles to provide the functionality required for autonomous operation. Many system-level operations involve measuring and managing power requirements to control power, overall consumption, and heat dissipation.

03 Five major challenges of autonomous vehicles

Today, after more than 50 years of continuous research and development, we see that self-driving cars have become a reality. Nevertheless, there are still many challenges in designing a fully autonomous system for self-driving cars.

1. Road conditions

Road conditions can be highly unpredictable and vary from place to place. In some cases, there are wide, flat, marked highways. In other cases, road conditions are severely degraded—with no lane markings. The same is true for undefined lanes, potholes, mountainous areas, and tunnel roads where external directional signals are not very clear.

2. Weather conditions

Weather conditions are another big drawback. There can be sunny days, but there can also be rainy and stormy days. Self-driving cars should be able to work in all weather conditions. There should be absolutely no room for glitches or downtime.

3. Traffic conditions

Self-driving cars have to drive on roads in all kinds of traffic conditions. They have to drive on the road with other self-driving cars and, at the same time, with many people. Wherever humans are involved, a lot of emotions are involved. Traffic can be highly moderated and self-managed. But there are often cases where people violate traffic rules. Objects may appear in unexpected situations. In dense traffic conditions, even a few centimeters per minute of movement is important. People cannot wait endlessly for traffic to clear automatically, and there are some prerequisites to start moving. If there are more such vehicles on the road waiting for traffic to clear, it may eventually lead to traffic gridlock.

4. Accident liability

The most important aspect of self-driving cars is accident liability. Who is responsible for an accident caused by a self-driving car? For self-driving cars, software will be the main component that drives the car and will make all the important decisions. While the initial design had a person behind the wheel, the new design shown by Google has no dashboard and steering wheel! In such a design, if the car does not have any controls like steering wheel, brake pedal, accelerator pedal, etc., then how are the people inside the car supposed to control the car in case of an unexpected accident? Moreover, due to the nature of self-driving cars, the passengers are mostly in a relaxed state and may not pay close attention to the traffic situation. In the event that a situation requires their attention, by the time they need to take action, it may be too late to avoid the situation.

5. Radar Jamming

Self-driving cars use lasers and radars to navigate. The lasers are mounted on the roof of the car, while the sensors are mounted on the body of the car. Radar works by detecting radio waves reflected from surrounding objects. When on the road, the car constantly emits radio frequency waves that reflect off surrounding cars and other objects near the road.

The time taken for the reflection is measured to calculate the distance between the car and the object. Appropriate action is then taken based on the radar reading. Radar works by detecting reflections of radio waves from surrounding objects.

When on the road, the car continuously emits RF waves which are reflected off surrounding cars and other objects near the road. The time taken for the reflection is measured to calculate the distance between the car and the object and then take appropriate action based on the radar reading.

When this technology is used with hundreds of vehicles on the road, will one vehicle be able to distinguish its own (reflected) signal from another vehicle's (reflected or transmitted) signal? Even if there are multiple radio frequencies available for radar, it is unlikely that this frequency range will not be available for all vehicles manufactured.

04 Conclusion of the Autonomous Vehicle Challenge

Even today, training algorithms to get self-driving cars on the road still presents many challenges. But so does the determination of scientists, engineers , and problem solvers from many different disciplines.

The industry's concerted efforts will surely make self-driving cars on the road a reality one day, and the benefits will be huge. Not only will it save fuel, encourage efficient transportation and shared services, but it will also help save the many lives that are often lost in road accidents.