One of the most rapidly advancing and exciting changes in present day society, that is simultaneously interwoven within the machine vision sector, can be found within the self-driving vehicle industry. In the near term, we can expect to see a boom in “robotaxis” and fleets of commercial trucks which will be fully autonomous. In the more distant future (about 8 years off) we can expect to see at least 10% of U.S. consumer vehicles operating on the road at a fully autonomous level (according to MicKinsey).
Some of the automakers on the forefront of this exciting frontier include Apple owned Waymo, GM’s Cruise, and Tesla. All of these autonomous cars have already logged millions of test-driving hours in cities across America.
But to fully realize such a future where the self-driving car is the new normal, the car must be, above all, safe and able to “see.” Currently, autonomous cars use a variety of tools such as lidar sensors and numerous cameras to allow it to “see” and react to what it is perceiving around it, as the two images below demonstrate.
The cameras mounted onto these vehicles use various car, pedestrian, and object detection algorithms (within software such as NVIDIA) to recognize the moving world around it as well as lanes, road signs, and other miscellaneous related surroundings.
At its onset, most car companies relied on standard USB and GigE industrial cameras for the computer vision component of the system, but recent trends have shown an increase interest in thermal cameras and 3D camera systems mounted onto these cars. Thermal has it’s benefits in the manner in which is doesn’t need illumination for imagery, thus it operates well whether in bright sunlight, complete darkness, or in any lighting glare. It also has increased awareness in harsh weather conditions, so overall, it offers a heightened level of safety to the autonomous ride. Stereo (3D) cameras have their benefits, in that they perceive depth in the environment, whereas a standard camera does not (and uses lidar to perceive a sense of depth). Additionally, some newer 3D cameras are able to scan up to hundreds of meters ahead and are also able to identify a wide variety of different elements and materials, such as ice on a road. These advancements are perhaps, why Elon Musk insists lidar “is a fool’s errand” and only uses cameras on Tesla vehicles.
Camera capabilities continue to evolve to fit the requirements of applications, market trends, and engineering demands. Although we are all witnessing the nascency of the self-driving industry, here at Lore Technology we are well equipped to assist through the ever-changing vision needs of automakers. Our various camera and lens manufactures have been playing a role in the vision supply pipeline for the autonomous vehicle industry for many years now. As such, we can offer excellent support and price points. Contact us today to see how we can help.
