In order to drive autonomously, vehicles need cognition (processes used to judge, plan, acquire knowledge, think, and etc). They must be able to understand what is around them. This context can be obtained in the form of a machine-readable, semantic map. Detailed 3D semantic maps, also commonly known as “HD maps,” have become the industry-wide standard for self-driving cars.
Even so, a 3D map is of no use to a vehicle without precise localization, the ability to position itself on the map. Like humans intending to go somewhere, the vehicle needs to know where it is before it can follow a path. Moreover, while this new generation of 3D maps is far more comprehensive, it is not actually sufficient for achieving Level Four (SAE) autonomous driving, where the human driver has no responsibilities towards control. Self-driving cars need much more in the form of cognitive tools to aid in decision-making.
To address this knowledge gap, we have developed techniques for localizing a vehicle in six degrees of freedom (6DoF). The concept video here shows the result of combining our highly detailed, 3D semantic map with localization in 6DoF (also referred to as 6D). Through localization in this manner, the car is given an additional layer of assistive map information. It can use this information to make smarter decisions and to drive safely. With both location and orientation from 6DoF, the vehicle can focus its sensors (foveation) towards a particular region in space, where a need-to-know action is occurring in the car’s environment.
For more information, please visit www.Civilmaps.com
Engineering blog: medium.com/@CivilMaps