Advances in sensor cleaning pave the way for autonomous driving

What happens to automated functionality when sensors are obscured by dead bugs or mud? Megan Lampinen explores ground-breaking cleaning technology to address the problem

Sensor systems provide pivotal environmental data to smart cars, allowing for automated and autonomous functionality. They can tell a vehicle exactly what lies on the road ahead and behind, as well as how the road itself is laid out.

But when those sensors become obscured by heavy rain, mud, snow, bird droppings or dead bugs, it causes serious hindrance to vehicle safety systems. Today, many new vehicles are forced to disable their active safety systems when the sensors become obscured. This is a real inconvenience for drivers, but it could prove a death knell for autonomous vehicles.

Ensuring sensor performance

The key is to clean these sensor surfaces in the timeliest and most effective manner. “Today’s vehicle safety systems are incredibly complex,” notes John Saxon, Chief Executive of fluid distribution specialist dlhBOWLES. “They provide vast amounts of data into the CPU, but none of it works if a bug hits the sensor.”

dlhBOWLES_system-disabled

Any sensor pointed outwardly is subject to environmental conditions. Something as common as heavy rain can prevent proper functioning of key safety systems. “This problem was amplified when the technology hit the market in more applications,” adds Saxon.

The number of vehicle sensors is growing rapidly, driven in part by new safety regulations. The US market rear camera mandate is a case in point. All new cars built since 1 May 2018 are required to have a rear-view camera, but there is no requirement to have a cleaning system for this camera. For cars without one, the driver may even need to exit the vehicle and manually clean the camera. That could jeopardise their safety, negating the benefits of these new systems. “Why would drivers be expected to get out of the car and maintain a safety system?” asks Saxon. “At the same time, God forbid there was ever an accident due to a dirty sensor that was compromised as a result of environmental factors.”

dlhBOWLES does not make safety systems; it aims to help safety systems operate effectively by keeping essential sensors free from environmental contaminants. The cleaning fluid system inevitably adds weight to the vehicle, so optimising fluid use will keep this addition to a minimum. The company has racked up five decades of experience in fluid distribution, but the approach that works for cleaning a windshield will not cut it with sensors. “We now face a new challenge,” says Saxon. “Zone cleaning is needed for cameras and sensors, as you don’t need to actuate all areas. It is not an option to simply install a bigger fluid tank, because real estate in the car has become more of an issue.”

A new partnership with German valve manufacturer RAPA targets this zone cleaning challenge specifically. RAPA’s solenoid distribution system allows dlhBOWLES to control when and how much air or liquid is distributed to a sensor. “The majority of what we do is air suspension, but we also have liquid dynamics based upon hydraulic applications,” explains RAPA’s Chief Executive Kelly Nelson. “The uniqueness of the partnership from a systems standpoint was having both.”

Originally the system was simply cleaning the sensors, but developers discovered that the residual liquid drops could distort the view. To solve the problem, nozzles and tubing were adapted to carry out air blow-off as well. Further developments are anticipated down the line. “We have come up with a really simple solution, but we are only in phase one,” says Nelson. “Over the next five years, as the market moves rapidly into autonomous driving, there will be challenges ahead. We are trying to anticipate the industry’s needs and have a solution ready for developers before they get there.”

dlhBOWLES dirty cars challenge sensors
“God forbid there was ever an accident due to a dirty sensor that was compromised as a result of environmental factors” – John Saxon, Chief Executive, dlhBOWLES

One of the hardest tasks at the moment is to meet future customer expectations. “With any new product, we can generally anticipate 80% of what’s needed, but 20% will be a challenge,” he adds. “What we are showing today is version 1.0, but we can see where versions 2.0 and 3.0 will come rapidly in succession.”

Regulatory gaps

Another significant challenge is defining exactly what is ‘dirty’ and what is ‘clean enough’. Different levels of dirtiness might be acceptable for different sensors at certain times. The aim is to clean a sensor with just the right amount of fluid at the moment just before it is needed. If a vehicle knows through GPS that it will turn 30 metres ahead, it could begin some preliminary checks just before that and determine if any of the required sensors need cleaning.

“We are working with automakers to define some of these standards,” Saxon tells Automotive World. “We have sat down with their autonomous vehicle (AV) teams to help determine specifications. At the moment there is no standard, so specifications are going to be all over the place. At some point, the government will need to look at the safety requirements for each level of automation and determine what’s required, and what can be retrofitted. There are many unanswered questions, but we are in a position to help answer them.”

Saxon suggests that consumers could prove a real driving force behind developments on this front, and that automakers will begin to specify cleaning systems even in advance of any regulatory requirements: “Consumers are becoming frustrated. Nobody likes receiving error messages that their systems are disabled, but many don’t know that cleaning systems exist. As awareness grows, demand will increase. We know these systems work. When that becomes a differentiator in the consumer choice, then we will hit the inflection point.”

Towards autonomy

The importance of clean sensors will only increase as vehicles move towards higher levels of automation. A wash and dry system like the one proposed by dlhBOWLES represents a simple solution. “We have millions of empirical data points showing how our systems work in all types of conditions,” says Saxon. “That leg work has been done. It means that automakers, LiDAR suppliers and software manufacturers can focus on what they need to, and leave the cleanliness of it to us.”

“Today’s vehicle safety systems are incredibly complex. They provide vast amounts of data into the CPU, but none of it works if a bug hits the sensor” – John Saxon, Chief Executive, dlhBOWLES

While the outlook for the automated driving timeline varies widely among industry players, Saxon anticipates steady progress in the years ahead. “No matter where it goes, it will continue to drive forward. We may not see fleets of AVs in the heartland of America next year, but they are coming.”

For Nelson, industry developments will hinge partly on the state of the economy: “If there’s an economic slowdown, some automakers will lose R&D support. That’s one of the first things to be cut. But we are in a position to still move them forward. Even though they can’t make investments, we can do the heavy lifting and the engineering side of things, so we will be ready when they come back, offering a solution they can incorporate quickly.”

Notably, the importance of strong partnerships will also grow as the industry faces steeper development and cost challenges along the road towards autonomy. “We are very good at what we do, but starting a process that duplicates what RAPA already brings would take years and unnecessary investment,” Saxon emphasises. “This partnership takes years off what it would have taken to develop on our own. We don’t have years. That’s why partnership is critical—even if you have the money, you don’t have the time. Development on your own is almost foolhardy.”

Close
Close