Autonomous Speed Planning for Path-Restricted Mobile Actors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous and semi-autonomous vehicles and robots restricted to predetermined paths face challenges in effectively planning their speed to avoid collisions with other objects, particularly in mixed environments where obstacles are unpredictable and emergency braking may not be sufficient.
Innovation Solution
A method for autonomous speed planning that uses sensor data to detect and classify objects, assign motion models, and determine a speed profile through a mixed-discrete-continuous optimization method, allowing for real-time adjustments to prevent collisions by anticipating the movement of other objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If emergency braking is used to avoid collisions, then safety is improved, but comfort deteriorates and productivity is reduced due to frequent stops
Solution Approach 1:
The system performs preliminary speed planning by detecting objects and predicting their future positions before collisions occur. The speed planner proactively adjusts the vehicle's speed profile in advance to avoid collisions, rather than reacting with emergency braking only when collisions are imminent. This preliminary action maintains safety while preserving comfort by avoiding sudden stops.
2Productivity
If the vehicle maintains constant speed to improve productivity, then efficiency is improved, but collision risk increases when objects are present on the path
Solution Approach 1:
The speed profile is made dynamic rather than constant. The system continuously detects objects, predicts their movements using motion models, and adjusts the speed profile in real-time based on the dynamic situation. This allows the vehicle to maintain high productivity when the path is clear while automatically reducing speed when objects are detected, thus avoiding collisions without unnecessary stops.
3Measurement precision
If the system detects and processes multiple objects with different motion models, then collision avoidance accuracy is improved, but computational complexity increases
Solution Approach 1:
The system segments objects into different categories (e.g., stationary objects, moving objects with different motion patterns) and assigns specific motion models to each category. This segmentation allows the use of simplified prediction models for each object type rather than applying a single complex model to all objects, thereby improving prediction accuracy while managing computational complexity through structured categorization.
Data Source
Figure 1
Figure 2~3
AI summary
A method for autonomous velocity planning of a moving actor (32) that is at least semi-autonomously controlled and restricted to a predetermined path is described. In this method, sensor data (SD) relating to the environment (U) of the moving actor (32) are acquired. Objects (O) in the environment (U) are also detected based on the acquired sensor data (SD). The detected objects (O) are categorized into different object classes (OK), with each object class (OK) being assigned at least one motion model (BM). Based on the at least one assigned motion model (BW), a future movement (ZB) and/or position (P) of the objects (O) is determined. Furthermore, a velocity profile (GP) of the moving actor (32), for which a predetermined risk criterion (RK) is met, is determined based on the future movement (ZB) and/or position (P) of the objects (O).A planning device (20) is also described. In addition, a movable actor, controlled at least semi-autonomously and restricted to a predetermined path, is described.