Simple Driverless Hardware Choices Yet to be Settled

Some simple but serious strategic decisions over hardware for driverless vehicles have yet to be properly settled.

That’s the opinion of Sandeep Sovani, director, global automotive industry, at Ansys when he told TU-Automotive about how connected and autonomous vehicles can overcome the challenges presented by different weather conditions. “From a hardware perspective this about making a strategic decision of which sensor to use; LiDAR, radar, cameras, etc.,” he explains before adding that the “second aspect comes as developing each sensor as a component that can see through the weather”.

Why is this important? Not all of the sensors have the same capabilities. While radar can see through a tick blanket of snow and rain, it can’t make out complex shapes. Cameras on the other hand would be blinded by these weather conditions. LiDAR and cameras don’t do as good a job as radar can in them. On the other hand, he says, LiDAR is better at picking out shapes but is short-ranged and is affected by weather conditions.

Each one uses a different frequency of the electro-magnetic spectrum at which they operate best during certain weather conditions. Cameras are limited when they can’t see. They require visible light to operate in the same way as human eyes do. “Whereas radar operates within millimeters, visible light is in the order of micro-meter,” he says before commenting that, “radar waves can walk around rain drop or snowflakes but visible light waves are obstructed by rain and snow.”

Sensor suite choice

So, the initial strategy for overcoming all weather conditions should consider the choice of which suite of sensors will cover all the bases. That includes a consideration of all the potential weather conditions that the vehicles will face. Sovani asks: “How many radars, what should be the range of the radars: short, medium or long-range? Questions should also be asked about the limitations of each type of sensor or device – including cameras and LiDAR.”

With a suite of sensors, it’s possible to gather all of the types of information required to keep connected and autonomous vehicles safe on the road. While the data is collated from radar, LiDAR and cameras, the information gleaned from them has to become fused to create a complete and comprehensive view of the world outside. Sovani says the software then needs to decide how much priority to give to each sensor. “If the car knows the weather is clear outside, then it can give a higher priority to camera but, if the weather is poor, it can give a harder priority to images coming from radar,” he explains.

All in the data  

He adds that connected and autonomous vehicles can know if the weather is good or bad by analyzing sensor output and by allowing them to continuously talk to weather networks and databases to work out the weather of a car’s current location or intended route: “Highway Pilot applications are talking through cellular networks to gain data about weather conditions. Under certain weather conditions the driver will get message stating that Highway Pilot is not available due to them.” This may mean that the driver has to take over control of the vehicle for safety reasons.

 

Important of connectivity

Without connectivity, the systems can’t work, says Sovani. He added: “Connectivity is important from the perspective that the car should be able to talk to weather stations to learn about them. As we go towards 5G communication, the amount of information that cars can have could be enormous. It is important when designing the networks to ensure that the infrastructure and equipment is capable of handling the weather. The 5G connections have to be reliable and so it’s essential to make sure they don’t get dropped no matter the weather.”

Cleanliness and maintenance

Beyond connectivity, there is the need to ensure that the sensors are clean, well maintained and fully operational. This means they should be designed, according to Sovani, to “operate well to high degree of capacity – even when there is some lack of clarity in front of the sensors”. One way to achieve this is by making sure that the sensors are regularly cleaned. The second aspect focuses on the hardware and software of the sensors, which need to be tolerant to the diverse range of weather conditions that the vehicle, or vehicles, could face. They also need to be able to tolerate dirt.

He explains: “Keeping the sensors clean includes having LiDARS with wipers to ensure that the visible surfaces are kept clean. Then there are some nice inventions, such as water repellents. We see examples: some sensors are put into headlights and they can have a camera and a LiDAR fitted; with the lenses of the headlights, there are companies developing special coatings so that the glass becomes water repellent.”

Road surface visibility

Just like human-beings need to see the road ahead to be able to analyze how to react to the obstacles and weather conditions around them when they are driving, or even when they are walking, there is a need for connected and autonomous vehicles to predict the condition of the road surface for ensuring their safe operation.  Impaired visibility could cause accidents.

Yet by knowing the condition of both the weather and the road surface ahead the vehicles can slow down or avoid making sharp turns if it slippery ahead. This ability is particularly important when the road surface is icy and slippery.

V2V communication

Sovani adds: “Vehicle-to-vehicle (V2V) communication can make autonomous vehicles superior to human-driven cars. This can communicate that there is an ice-patch ahead to other vehicles to allow them all to take corrective action automatically.  Autonomous vehicles don’t only have sensors; they have a telepathic sensor to communicate with other vehicles to relay information about road surface conditions.”

He concludes by arguing that the technology is heading in the right direction, which entails the creation of increasingly robust algorithms. “This is important from a safety perspective, making it possible to make the right decisions about objects around the vehicle based on weather and lighting conditions. There also radars and LiDARs that are being tested and developed to be more resistant to weather. Different LiDARs are being tried out, and MIMO radars are being worked on as they will be more robust for general applications, etc.”

Importance of simulations

Ansys provides engineering simulations to assist with sensor and component development. Testing them without simulation might require carmakers to wait for months for snow, rain, fog, or for another types of weather condition. With a simulation it’s possible to allow manufacturers to test their products in a much faster and more economic way.

“Autonomous vehicle manufacturers have realized that the amount of road or physical tests to ensure that safety of autonomous vehicles would otherwise take centuries to complete and so it is impossible to develop autonomous vehicles without simulation.” Simulation, therefore, has to be part of the armory for overcoming all weather conditions – along with the choice of sensor and device suite for enabling the vehicle to see, analyze and react appropriately as the weather changes.


Leave a comment

Your email address will not be published. Required fields are marked *