Autonomous Vehicle Path Planning Using Spatial MPC
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Legacy sampling-based path planning systems for autonomous vehicles face inefficiencies when vehicle velocity is low or approaching zero and encounter difficulties in space-constrained environments, such as those with other vehicles.
Innovation Solution
A path planning system that combines a constraints computation system and a spatial model predictive control (MPC) system to apply constraints and optimize paths, comparing potential paths to select the most suitable one for navigation, using both velocity-based and spatially-constrained approaches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a sampling-based approach is used for path planning, then the system can identify paths in general environments, but it becomes inefficient when vehicle velocity is very low or approaching zero
Solution Approach 1:
The path planning system dynamically adapts its approach based on vehicle velocity. When velocity is high, it uses sampling-based planning; when velocity approaches zero, it switches to model predictive control with spatial optimization. This dynamic adaptation resolves the contradiction by adjusting the planning methodology to match the operational conditions.
Solution Approach 2:
The system changes the planning parameters and time horizon based on velocity conditions. At low velocities, it extends the spatial planning horizon and adjusts the optimization parameters to account for the reduced temporal component, allowing efficient path planning regardless of vehicle speed.
2Adaptability or versatility
If a sampling-based approach is used for path planning, then the system can operate in various environments, but it encounters difficulties when paths are constricted by the presence of other vehicles
Solution Approach 1:
The path planning is segmented into two distinct phases: sampling-based reference path generation and model predictive control optimization. The MPC phase specifically handles constricted environments by optimizing the trajectory within the reference path framework, making the system more effective in narrow or constrained spaces while maintaining overall environmental adaptability.
3Stability of the object's composition
If the time horizon shrinks at low velocity, then the vehicle can maintain stability, but the path planning system becomes unable to efficiently produce proposed paths beyond that time horizon
Solution Approach 1:
The system transitions from purely temporal planning to spatial planning by formulating the optimization in terms of spatial distance rather than time. This dimensional change allows the planner to extend beyond the restricted time horizon while maintaining stability, as the spatial formulation naturally accounts for the vehicle's state without being constrained by the shrinking time window.
Data Source
AI summary
Among other things, techniques are described for identifying, by at least one processor of a vehicle based on a graph that includes a plurality of edges and a plurality of nodes, a reference path through an environment that includes a subset of the plurality of edges. The technique further includes identifying a first path based on optimization of a spatial model related to the graph and the reference path and a second path based on application of at least one constraint to the reference path. The technique further includes selecting, by the at least one processor, the first path or the second path as a path along which the vehicle will traverse based on a pre-identified rulebook. Other embodiments may be described or claimed.