Autonomous Vehicle Trajectory Planning for Critical Segments
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory planning techniques for autonomous ground vehicles (AGVs) are limited in navigating critical scenarios such as sharp bends, twisting paths, reversing, and navigating through unorganized spaces like factories or ports, especially when encountering other vehicles or limited parking areas.
Innovation Solution
A method and system that assesses the need for a trajectory plan for critical segments of a path, determines a next free road region using an AI model based on current and goal poses, vehicle capabilities, and generates a trajectory sub-plan by iteratively determining intermediate poses between rest poses to reach the final goal pose.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing trajectory planning techniques are used, then the system is simple to implement, but the AGV cannot navigate critical scenarios such as sharp bends, twisting paths, and reversing
Solution Approach 1:
The trajectory planning is divided into multiple sub-plans, each handling specific critical scenarios (sharp bends, twisting paths, reversing). The system segments the overall navigation task into manageable parts that can be addressed by specialized planning algorithms for each scenario type.
Solution Approach 2:
The trajectory planning system dynamically adapts its approach based on the detected scenario type. When critical scenarios are identified through sensor data analysis, the system switches from standard trajectory planning to scenario-specific planning strategies, making the system flexible and adaptive to varying navigation conditions.
2Adaptability or versatility
If existing trajectory planning techniques are used, then the computational requirements are low, but the AGV fails to handle difficult parking scenarios with limited space
Solution Approach 1:
The system performs preliminary assessment of the parking scenario by analyzing sensor data to detect limited space conditions before attempting maneuvering. Trajectory sub-plans are pre-computed for various parking scenarios, allowing the AGV to select and execute appropriate maneuvers without real-time computational overhead during critical moments.
Solution Approach 2:
An intermediary assessment module analyzes sensor data and scenario conditions to determine when difficult parking scenarios are present. This intermediary layer translates complex sensor inputs into scenario classifications, enabling the trajectory planning system to activate specialized parking maneuvers only when needed, balancing computational efficiency with capability.
3Adaptability or versatility
If existing trajectory planning techniques are used, then the system responds quickly to changes, but the AGV cannot navigate unorganized spaces like factories or ports where loading and unloading occur
Solution Approach 1:
The system changes its planning parameters and constraints based on the detected environment type. In unorganized spaces like factories or ports, the trajectory planning algorithm adjusts parameters such as obstacle avoidance margins, speed limits, and path smoothness requirements to accommodate dynamic loading and unloading activities while maintaining navigation capability.
Data Source
AI summary
The disclosure relates to method and system for generating trajectory plan for autonomous ground vehicles (AGV). The method includes assessing an upcoming goal position for an AGV with respect to a current position of the AGV to establish a need for preparing a trajectory plan for a critical segment of the path. The method further includes determining, upon establishing the need, a next free road region based on a hint next rest pose along the critical segment, generating a trajectory sub-plan corresponding to the next free road region by iteratively determining a set of intermediate poses between the current rest pose and the hint next rest pose, and generating the trajectory plan for the critical segment of the path by iteratively generating the trajectory sub-plans between the current rest pose and the final goal pose.