Maintaining Clear Vision for Autonomous Trucks

By TuSimple's editorial team
Mar 26


Every driver has been in the unwelcome situation of having an obstructed view. Maybe you parked under a tree to keep the car cool while you went to work only to find your car covered in pollen and leaves when you return. Maybe you were unfortunate enough to pass through a swarm of insects at highway speeds leaving unsightly splatter across your windshield. Maybe the whole car was draped in frost on a wintry morning. Whatever the reason, obscured vision is frustrating and potentially dangerous.

As a human driver, you can spray the windshield with cleaner, activate the wipers, or turn on the defrosters. Some cars can detect moisture and activate or adjust the wipers automatically. Just as you rely on your vision to navigate a vehicle safely, an autonomous vehicle relies on an array of highly sensitive sensors working together to create a comprehensive image of the road and conditions ahead. Each device is precisely calibrated and positioned to provide complete and consistent data. When one or more sensors are obstructed, however, the resulting data might not provide sufficient information for the vehicle to navigate safely and efficiently.

Adverse conditions for the sensors can range from routine weather like rain or snow to blowing dust and small particles, bugs, and even bird droppings or splashes from other vehicles. When precipitation or debris interfere with the clarity of the camera images or the sensor readings, they can degrade the performance of autonomous vehicles. Naturally, our redundant sensors are available to back up any obstructed sensors, but the ability to automatically clean each sensor is vital to efficient operation. Otherwise, additional time and resources must then be spent cleaning the affected devices until they are back within an acceptable range of function and the vehicle can return to normal operation.

We are often asked about the solutions we have developed for problems like this. These issues have always been part of the comprehensive approach to our autonomous fleet. Consequently, we have been working on a variety of solutions that are represented in our growing collection of patent assets related to camera and sensor cleaning technologies. These patent assets demonstrate our industry leadership in these technologies just as our overall patent portfolio demonstrates our leadership in autonomous trucking as a whole.

Just as there are distinct types of debris and conditions that affect sensors, we have considered that there are many different ways of cleaning them. These designs reflect our thoughtful development of systems to identify and activate appropriate cleaning techniques for each potential situation.

We have solutions running across an array of sensor cleaning configurations. These include alternate movement of the wiper systems, varying intervals for wiper activation, multiple intensities of fluid delivery, designs based on weather-resistant materials, and processors to control different sensor cleaning configurations. For one example of these solutions, see our US Patent Application No. 17/363887, which describes a platform including multiple cameras, lateral wiper systems, and a fluid delivery system as well as processors capable of controlling each system. Our US Patent No 11,066,048 describes an autonomous cleaning mechanism for sensors including a housing, pivoting wipers, jets for additional cleaning, and a heating element for the protective sensor cover. We have also considered how distinct types of cleaning can be performed autonomously based on different types of sensors and camera obstructions. For example, our US Patent Application No. 16/194048 describes a system to clean a protective cover of a camera based on input from both the camera and external sensors. Such a system can include a control unit to determine the type of obstruction including both solids and liquids of varying viscosity, select an optimal cleaning method, and initiate the cleaning.

These diverse solutions show the depth of commitment we have made to developing unique and adaptable innovations to both sensor issues and the industry in general. Automated cleaning solutions can extend the effective operation of sensors as well as their ability to function optimally in varying conditions. We don’t plan on stopping here. We are consistently growing our patent portfolio on detecting and resolving adverse travel conditions. We are creating a better path forward with practical technology and intellectual property that continually advances the field of autonomous freight.

In the fourth quarter of 2021, an additional 30 new patents were awarded to TuSimple, giving us a total of 387 patents. These latest patents cover essential technologies, such as simulation systems, map construction, and perception housing innovations for variable weather condition operation efficiency. They join an ever-growing portfolio of patents representing a broad range of technologies providing real-world solutions for autonomous freight. Our goal is to ensure the highest quality freight network while continuing to lead the industry in safety, reliability, efficiency, and sustainability.

Posted by TuSimple

Recent Blog Posts