Vehicle Trajectory Planning Under Dynamic Geometric Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory planning methods for autonomous driving systems struggle to efficiently handle complex nonlinear and dynamically changing constraints, leading to instability and safety issues in vehicle navigation.
Innovation Solution
A vehicle trajectory planning method that incorporates geometric and dynamics constraints, using reinforcement learning and iterative linear quadratic regulator (iLQR) to optimize trajectories, considering energy loss, acceleration, and angular speed constraints, ensuring smooth and safe navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing trajectory planning methods are used, then the system can handle basic navigation, but the system becomes unstable and unsafe when facing complex nonlinear and dynamically changing constraints
Solution Approach 1:
The patent applies dynamics by making the trajectory planning system adaptive to changing conditions. The cost function dynamically adjusts weights for different constraints (energy loss, acceleration, angular speed) based on real-time vehicle state and environmental conditions. This allows the system to handle complex nonlinear and dynamically changing constraints while maintaining stability and safety through continuous optimization.
2Reliability
If the trajectory planning considers multiple constraints (energy loss, acceleration, angular speed), then the navigation becomes safer and smoother, but the computational complexity increases
Solution Approach 1:
The patent changes parameters by formulating a cost function that incorporates multiple constraints (energy loss, acceleration, angular speed) with adjustable weights. The optimization algorithm iteratively adjusts these parameters to find the optimal trajectory that balances safety, smoothness, and computational efficiency. This allows the system to consider multiple constraints without excessive computational burden.
3Ease of operation
If the trajectory planning considers multiple constraints (energy loss, acceleration, angular speed), then the navigation becomes smoother and more comfortable, but the energy consumption for computation increases
Solution Approach 1:
The patent applies partial action by selectively emphasizing certain constraints based on the current driving situation. The cost function includes terms for energy loss, acceleration, and angular speed with adjustable weights that can be tuned to prioritize smoothness and comfort when conditions allow, while reducing computational energy consumption by not always maximizing all constraint considerations simultaneously.
Data Source
AI summary
A vehicle trajectory is planned. An initial reference trajectory of a target vehicle within a target planning duration is acquired. The initial reference trajectory includes an initial state variable and an initial control variable of the target vehicle at at least one position point. A reference lane trajectory is acquired. A trajectory cost of the target vehicle is determined according to a geometric constraint and a dynamics constraint by using the reference lane trajectory and the initial state variable and the initial control variable of the target vehicle at the at least one position point. A target vehicle is controlled to adjust from the initial reference trajectory of the target vehicle to a target travelling trajectory according to the trajectory cost of the target vehicle.