Autonomous Vehicle Trajectory Planning via Maneuver Pattern Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current trajectory planning systems for autonomous and semi-autonomous vehicles are computationally intensive, making them slow and inefficient for real-time reactive applications, which affects the safety and comfort of on-road driving.
Innovation Solution
A method that ranks maneuver patterns based on comfort scores and local parameters, allowing for the selection of a maneuver pattern that meets predetermined safety and comfort thresholds, with the option to adjust thresholds if initial selections do not meet requirements, thereby optimizing trajectory planning with reduced computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If typical path planning techniques are used to ensure safe and comfortable trajectories, then trajectory safety and comfort are improved, but computational time increases making real-time reactive applications infeasible
Solution Approach 1:
The patent segments the trajectory planning process into two distinct phases: (1) a fast reactive phase that generates candidate trajectories using simplified models for immediate response to dynamic changes, and (2) a slower optimization phase that refines selected candidates. This segmentation allows the system to meet real-time deadlines while still achieving safe and comfortable trajectories through iterative refinement of pre-selected candidates.
Solution Approach 2:
The patent performs preliminary action by pre-computing and storing optimal trajectory segments offline that satisfy safety and comfort constraints. During real-time operation, the system retrieves and assembles these pre-computed segments based on current conditions, avoiding the need to perform full trajectory optimization in real-time while maintaining trajectory quality.
2Reliability
If computationally intensive path planning techniques are used to optimize trajectory comfort and safety, then trajectory quality is improved, but system responsiveness deteriorates
Solution Approach 1:
The patent implements dynamic adjustment of planning effort based on operational context. The system uses a two-level architecture where a fast reactive planner provides immediate responses using simplified dynamics models, and a slower optimal controller refines trajectories when computational resources are available and conditions permit. This dynamic allocation of computational effort maintains responsiveness while achieving high trajectory quality when possible.
Solution Approach 2:
The patent changes key parameters of the trajectory planning system based on operational needs. The system adjusts the level of optimization detail, time horizon, and computational resources allocated to trajectory planning based on factors such as vehicle speed, environmental complexity, and urgency of maneuver requirements, enabling flexible trade-offs between responsiveness and trajectory quality.
3Speed
If real-time reactive trajectory planning is implemented to respond quickly to changes, then system responsiveness is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational workload into distinct modules with different complexity levels: a fast reactive planner using simplified models for immediate response, and a slower optimal controller using full dynamic models for refinement. This segmentation allows the system to maintain low computational complexity for time-critical operations while performing more complex calculations when time permits.
Solution Approach 2:
The patent substitutes complex mechanical computation with lookup tables and pre-computed trajectory segments stored in memory. Instead of performing full dynamic optimization in real-time, the system retrieves pre-calculated optimal trajectories based on current state, dramatically reducing real-time computational complexity while maintaining trajectory quality.
Data Source
AI summary
Technical solutions are described for maneuvering a vehicle autonomously is described. An example method includes receiving, by a controller, a command to navigate the vehicle, the command based on a global route calculation. The method further includes ranking, by the controller, maneuver patterns from a list of maneuver patterns to generate a ranked list of maneuver patterns. The method further includes selecting a maneuver pattern from the ranked list, and in response to a comfort score assigned to the selected maneuver pattern being above a predetermined threshold, executing the maneuver pattern.