Lebesgue-state-based Behavior and Motion Planning


Reference #: 01395

The University of South Carolina is offering licensing opportunities for Lebesgue-state-based Behavior and Motion Planning


In an autonomous driving system (ADS), the behavior and motion planning systems are responsible for generating a motion trajectory reference for the motion controller. This generated trajectory needs to be collision-free and comfortable, while taking an efficient path. To achieve these goals, the behavior and motion planning has to have a high-resolution motion trajectory. However, a high trajectory resolution comes at a computational cost in the ADS. currently, the computation resolution is based on a fixed time interval or distance, considering the tradeoff between the high resolution and computational cost. A method that can keep both high waypoint resolution and low computation cost is needed for an efficient automated driving system. 

Invention Description:

Lebesgue’s Theorem states that the value of an integrable function can be estimated at any point using infinitesimal averages around such point. By applying this, the generated waypoints of the ADS do not have a fixed time or distance interval. The varying distance of each waypoints depends on the curvature of the path. In a more curved path, more reference points are needed in order to effectively estimate the path that needs to be taken, in straighter paths, less reference is needed, and thus less computing

Potential Applications:

This fundamental idea can be applied in any use of automated vehicle operation. Tesla, Uber, Lyft, and OE auto manufacturer, or any organization or business interesting in self-driving capability can utilize this technology to optimize computing and allow for more precise automated driving using less computing power at any given point. 

Advantages and Benefits:

The advantage of Lebesgue-state-based behavior and motion planning is that it generates high resolution waypoints when vehicle is planning to do a lane change or turning operation and generates low resolution waypoints when vehicle is planning to do a simple straight lane-keeping drive operation. This decreases the computational workload of autonomous driving system, allowing the technology to effectively be used with less demanding resources. It also allows for scaling of the precision of the car’s placement, which can be especially useful in low speed, high accuracy situations such as parking.

Patent Information:
For Information, Contact:
Technology Commercialization
University of South Carolina
Bin Zhang
Zhichao Liu
Kai Zhang
© 2022. All Rights Reserved. Powered by Inteum