Robot Motion Planning With Neural Trajectory Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motion planning algorithms for autonomous systems are computationally expensive, especially when considering multiple degrees of freedom, leading to tradeoffs in accuracy and increased computation time, making them impractical for real-time applications.
Innovation Solution
A deep learning-based approach that uses a neural network trained on trajectories generated by a non-convex optimizer to rapidly compute motion trajectories, incorporating kinematic and dynamic limits, and jerk constraints to produce smooth and efficient motion plans, with a sequential quadratic program (SQP) used to refine the trajectories and ensure feasibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional motion planning algorithms are used to compute trajectories considering all constraints, then motion planning accuracy is improved, but computation time increases significantly
Solution Approach 1:
The system performs preliminary action by pre-computing and storing optimal trajectories in a database before real-time operation. During actual motion planning, the system queries the pre-computed database rather than calculating trajectories from scratch, dramatically reducing computation time while maintaining accuracy.
Solution Approach 2:
The system creates copies of optimal trajectories by storing them in a database for future reuse. Instead of重新computing trajectories for similar situations, the system retrieves and adapts pre-computed trajectory copies, significantly reducing real-time computation requirements.
2Adaptability or versatility
If additional degrees of freedom are introduced in the autonomous system, then system versatility is improved, but computation time increases
Solution Approach 1:
The system segments the motion planning problem by handling each degree of freedom independently through separate controllers and pre-computed trajectory components. This modular approach allows the system to manage multiple degrees of freedom without exponentially increasing computation time, as each segment can be planned and executed semi-independently.
3Productivity
If tradeoffs in accuracy are made to reduce computation time, then computation speed is improved, but motion planning accuracy deteriorates
Solution Approach 1:
The system performs preliminary computation of accurate trajectories and stores them in advance. During real-time operation, it retrieves these pre-computed accurate trajectories rather than computing them on-the-fly, achieving both high speed and high accuracy simultaneously.
Solution Approach 2:
The system creates and stores copies of accurate trajectory solutions in a database for future retrieval. This allows the system to maintain high computation speed by reusing pre-computed accurate trajectories rather than sacrificing accuracy for speed.
Data Source
AI summary
Methods and systems are provided for high-speed constrained motion planning. In one embodiment, a method includes computing, with a neural network trained on trajectories generated by a non-convex optimizer, a trajectory from one or more initial states of an autonomous system to one or more final states of the autonomous system, updating, with the non-convex optimizer, the trajectory according to kinematic limits and dynamic limits of the autonomous system to obtain a final trajectory, and automatically controlling the autonomous system from an initial state of the one or more initial states to a final state of the one or more final states according to the final trajectory. In this way, efficient and smooth trajectories can be rapidly computed for effective real-time control while accounting for obstacles and physical constraints of an autonomous system.