Trajectory Planning for Autonomous Vehicles Using Kinematic Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 2D path generators for motor vehicles fail to ensure that the selected trajectory is physically feasible and can be accurately followed by the vehicle, particularly for autonomously driving vehicles, as they do not account for the vehicle's actuators' capabilities in making necessary direction and acceleration changes, leading to potential issues with maintaining the trajectory.
Innovation Solution
A trajectory planning method that includes providing initial and end position sets, control sets for the vehicle, and a vehicle motion modeling function to determine a control profile, which considers the kinematic motion model and solver to generate a control profile indicative of the trajectory, ensuring the vehicle can physically follow the path.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current 2D path generators use a cost function to select trajectories from a group of trajectories, then a best possible path can be determined, but it cannot be ensured that the motor vehicle has the physical capabilities to drive along this trajectory
Solution Approach 1:
The patent applies preliminary action by incorporating the vehicle's kinematic motion model into the trajectory planning process before trajectory execution. The solver uses the vehicle model to predict whether the vehicle can physically follow the selected trajectory, checking capabilities such as direction changes and acceleration limits. This preliminary verification ensures that only physically feasible trajectories are selected, resolving the contradiction between trajectory optimality and physical feasibility.
2Device complexity
If the trajectory is determined without considering actuator control capabilities, then computational complexity is reduced, but the resulting end states of actuators may not be suitable for continuing movement
Solution Approach 1:
The patent implements feedback by using the vehicle motion model to predict actuator states along the entire trajectory, not just at the endpoint. The solver continuously checks whether the predicted actuator states remain within feasible ranges and whether the final actuator states are suitable for continuing movement. This feedback mechanism ensures reliable actuator states without requiring excessively complex computational models.
3Ease of operation
If the trajectory planning does not account for kinematic motion model, then the planning process is simpler, but the vehicle cannot accurately follow the determined trajectory
Solution Approach 1:
The patent applies parameter changes by incorporating the vehicle's kinematic motion model parameters (such as maximum steering angle, maximum acceleration, and other vehicle-specific characteristics) into the trajectory planning process. The solver uses these parameters to generate trajectories that are tailored to the specific vehicle's capabilities, ensuring accurate trajectory following while maintaining computational efficiency through optimized solving algorithms.
Data Source
Figure 1
Figure 2
Figure 3(a)~3(i)
AI summary
The invention relates to a trajectory planning method for ascertaining a control profile for a motor vehicle, in particular for a self-driving motor vehicle, said method having the following steps: providing a first position set (10) and a second position set (20), said first position set (10) having at least one piece of information relating to a starting condition at a trajectory start and said second position set (20) having at least one piece of information relating to an end condition at a trajectory end; providing a first and a second control set (30, 40), said first control set (30) having at least one piece of information relating to a first control state of the motor vehicle (100) at the trajectory start and said second control set (40) having at least one piece of information relating to a second control state of the motor vehicle at the trajectory end; providing a vehicle movement modeling function (50) which has a kinematic movement model of the motor vehicle (100); providing a solver (60); ascertaining a control profile information set (70), which has two profile information values (71, 72) that are indicative of the trajectory (200), by using the solver (60) on the vehicle movement modeling function (50) dependent on the first and second position set (10, 20) and the first and second control set (30, 40); and determining a control profile (110) for the motor vehicle (100) dependent on the control profile information set.