Ford and Baidu lead $150M investment in Velodyne lidar for self-driving cars
Ford and Chinese Internet search provider Baidu will each invest $75 million in Velodyne for automotive lidar.
IMAGE: Ford CEO Mark Fields shows off the Velodyne Puck sensor at a press conference on CES Press Day, January 5, 2016 in Las Vegas, NV ahead of the CES 2016 Consumer Electronics Show. The new Velodyne lidar models will be fitted into the side view mirrors of Ford’s Fusion Hybrid self-driving cars. (Image credit: Robyn Beck/AFP/Getty Images)
Ford (Dearborn, MI) and Chinese Internet search provider Baidu (Beijing, China) will each invest $75 million in Velodyne (Morgan Hill, CA) to help the company commercialize its laser-based light detection and ranging (lidar) sensors.
RELATED ARTICLE: Lidar nears ubiquity as miniature systems proliferate
In the early years of autonomous vehicle development, most of the companies involved were using Velodyne’s sensors that initially cost $80,000 to $100,000 or more. However, the company’s latest generation sensor--which is about the size of two hockey pucks stacked up--is now down to $8,000. Before the technology can be widely deployed in production the cost will have to come down to no more than a few hundred dollars.
Velodyne, like other companies developing lidar sensors, is moving away from the complex spinning laser toward solid state sensors that have no moving parts and should reduce the cost and make it more reliable and durable when subjected to the vibrations and temperature variations faced by vehicles.
Another Silicon Valley startup, Quanergy Systems, is currently seen as a leader in development of solid-state lidar and expects to have a production sensor for $250 in the 2018-19 time frame although that is not a 360-degree sensor. Four or more such sensors will be needed to provide a full surround view. While most companies in the auto industry view lidar as one of the key enabling technologies to bringing fully autonomous driving capability to the road, the most notable exception is Tesla where CEO Elon Musk has roundly dismissed the technology and called it unnecessary.
Scanning Lidar has the ability to generate a detailed point cloud of all the objects around a vehicle and in some cases can even recognize certain types of targets such as vehicles and pedestrians. Earlier this year, Ford took advantage of the lidar on its autonomous Fusion test vehicles to conduct the first autonomous tests on snow-covered roads. Prior to the snowfall, the automaker had generated a detailed 3D map of the test facility in Ann Arbor, MI and then used the lidar to locate the vehicle in space by recognizing previously mapped landmarks such as signs, trees, and buildings.
A system such as Tesla's AutoPilot that relies solely on optical cameras, radar, and ultrasonic sensors would not be able function in this type of environment where the road is not visible.
However, until now lidar has not performed well when there is rain or snow falling. The laser beams would reflect off individual rain drops and snowflakes, generating noise in the image of the vehicle’s surroundings. Lidar manufacturers are developing digital filtering algorithms to minimize this effect and that technology will be needed before autonomous driving can be broadly deployed.
Another problem that will need to be resolved before deployment is how to keep all of the sensors clean at all times. When driving in poor conditions, radar, cameras and lidar can all get covered in road residue that reduces or eliminates their ability to see. If autonomous vehicles are ever to be deployed for on-demand mobility services, they will need to function under all conditions, something they cannot yet do.
In addition to this funding news, Ford also announced directly that it is embedding engineers in the robotics lab at the University of Michigan (Ann Arbor, MI) to accelerate autonomous vehicle research.