19Feb - March 2021 that it is a stationary object (rather than something like a small, moving animal).In this situation, the vehicle not only needs to detect the object but also classify it as a tire fragment which must be avoided. Then the car must determine the right course of action, such as to change lanes to avoid the tire fragment and other incoming vehicle or object. For the car to have enough time to change its path and speed, these steps must all happen in less than a second. Again, these decisions made by the vehicle's onboard computer depend on accurate data provided by the vehicle's sensors.A Closer Look at Sensor TechnologiesTo be categorized as "fully autonomous," a car must be able to navigate between destinations without any intervention from a human driver. Self-driving cars aim to increase safety by eliminating human errors from driving situations, such as cell phone distractions or drowsy inattention.Sensor technologies perceive a car's environment and provide the onboard map with information about current roadway conditions. To build redundancy into self-driving systems, automakers utilize an array of sensors, including cameras, radar, and lidar. Camera-centric sensor suites can monitor the environment and enable limited driving automation. Cameras can identify colors and fonts, so they are capable of reading traffic signals, road signs, and lane markings. However, images produced by cameras ­ even stereo cams ­ do not always provide the level of accurate depth perception necessary for full autonomy. Due to the limitations in camera technology, we are yet to achieve complete autonomy or "self-driving" capability in the real world. Radar systems complement cameras very well. They typically offer better range and horizontal field of view. Radars are unhampered by inclement weather or lack of light. Additionally, radars provide accurate information on speeds of other vehicles. That said, radars have poor resolution (>10+ cm), so a radar's 3D image is unacceptably fuzzy. Radars also have difficulty detecting stationary objects. Consequently for accurate object detection and classification radars cannot be used without combining them with cameras.Lidar provides high-resolution, three-dimensional information about the surrounding environment. Unlike radar, lidar offers much higher resolution computer perception data, enabling accurate object detection. Unlike cameras, lidar provides accurate depth perception, with distance accuracy of a few centimeters, making it possible to precisely localize the position of the vehicle on the road and detect available free-space for the car to navigate. Lidars also offers 360 degrees horizontal field of view and up to 40 degrees vertical field of view, providing the vehicle the ability to generate dense, high-resolution 3D maps of the environment.Autonomous vehicles depend on the data provided by their sensors to perceive and navigate the environment. AVs will be equipped with lidar, cameras, and radar to enable safe, reliable full autonomy. Due to the limitations in camera technology, we are yet to achieve complete autonomy or "self-driving" capability in the real worldAnand Gopalan
< Page 9 | Page 11 >