Challenges
Save water
The most commonly seen AV sensor cleaning system is bump, nuzzles and a cleaning liquid system, in which the liquid is shared with the vehicle window cleaning tank. And mostly the vehicle computer set priority of water usage to the windshield cleaning function. Thus the AV sensor, including camera and lidar modules would get approx. 30% volume of the water reservoir.
Time of Activation
So to make the best usage of the limited liquid reservoir would be the first consideration when designing a cleaning strategy. By achieving this target, the cleaning sequence and timing need to be carefully planned after a series of field tests. The activation time of the nuzzle, gap of the activation time, and pump power, nozzle design parameters, etc. all need to be considered. The activation time is a tricky parameter that if it is too long, it brings up the water consumption, and also blocks the sensor during the operation of cleaning, if it is too short, the blockage wouldn’t be cleaning enough until the quality and quantity of information that sensor acquired is maintained.
Type of Blockage
The type of blockage along the year varies much, like dirt, mud, insects, snow, ice, and raindrops etc. The physical properties of each type are so different that only one setting of cleaning sequency wouldn’t work at an optimal condition.
Possible Solutions
In order to maintain good sensitivity of the sensor of an autonomous vehicle while executing a manual or an autonomous driving mode, a cost effective cleaning system is necessary to be mounted on the autonomous driving vehicle. It is preferable that the cleaning system would provide sufficient cleaning while consuming less reservoir. Due to the weather and environmental variation while driving, the blockage and residue on the front window of the lidar sensor would differ, so that the degradation of lidar performance varies with type of blockage and residues. The consumption and cleaning sequence, pattern or method varies under different types of blockage or residues. For example, under the rainy weather, while the vehicle is driving at a certain speed, the water drop stays on the front window if only utilizing the liquid cleaning system.
A combined cleaning system for the lidar sensor of autonomous vehicles is proposed with an air cleaning unit and a liquid unit, with these two units operating together based on a carefully designed cleaning strategy, which is according to different weather/road/environment conditions. The cleaning efficiency can be improved and at the same time save the consumption of water.
Summary
The combined cleaning solution provides a wider coverage under different use cases of an autonomous vehicle compared with a cleaning system with liquid only. Here the use cases include worse weather scenario, road conditions. It offers the opportunity to save liquid consumption and improve cleaning efficiency, while the air cleaning action can support the completion of a cleaning cycle better.
Comments