Automated Vehicle Trajectory Planning Using Occupancy Grid Reliability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Unpredictable and unknown scenarios pose a risk to the safety of at least partially automated vehicles, as existing systems struggle to accurately represent and respond to all possible environmental conditions.
Innovation Solution
The method employs occupancy grids to represent the vehicle environment, using a machine learning model trained to predict occupancy at a subsequent time based on historical data. This model generates a reliability measure by comparing predicted and actual occupancy data, allowing for informed trajectory planning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If occupancy grids are used to represent the environment, then the complexity and susceptibility to errors of object formation is eliminated, but the ability to represent unpredictable and unknown scenarios remains limited
Solution Approach 1:
The system implements feedback by comparing the predicted occupancy grid from the machine learning model with the actual occupancy grid from sensor data. This comparison generates a reliability measure that feeds back into the trajectory planning process, allowing the system to adjust its behavior based on the accuracy of its predictions in real-time.
Solution Approach 2:
The machine learning model performs preliminary action by predicting the future occupancy grid before the vehicle actually moves. This prediction allows the system to assess potential scenarios and calculate reliability measures in advance, enabling proactive safety adjustments rather than reactive responses.
2Loss of time
If a machine learning model predicts occupancy based on historical data, then trajectory planning can account for future states, but the model cannot represent all eventualities
Solution Approach 1:
The system uses multiple short-term predictions from the machine learning model, each valid only for the immediate future horizon. Rather than relying on a single comprehensive model that must cover all possible scenarios, the system generates successive predictions for near-future states, discarding older predictions as new data becomes available. This allows the system to adapt to new information continuously.
Solution Approach 2:
The system changes parameters by adjusting the prediction horizon and history length based on the calculated reliability measure. When reliability is high, the system can plan further ahead; when reliability decreases, the system shortens the prediction horizon and relies more on real-time sensor data, thereby adapting to the quality of available information.
3Reliability
If the system mitigates unknown situations by purposely braking, then safety is improved, but productivity and driving efficiency decrease
Solution Approach 1:
The reliability measure is calculated locally for different regions of the occupancy grid rather than as a single global value. This allows the trajectory planning to adjust locally in areas of low reliability while maintaining normal operation in areas of high reliability, thereby preserving driving efficiency while enhancing safety where needed.
Solution Approach 2:
Instead of always applying full mitigation actions like purposive braking, the system applies partial actions proportional to the calculated reliability measure. When reliability is moderate, the system makes subtle trajectory adjustments rather than abrupt braking, maintaining efficiency while still addressing safety concerns. Only when reliability is very low does the system apply excessive action such as purposive braking.
Data Source
AI summary
A method for planning a trajectory for an at least partially automated vehicle. The method includes: providing a trained machine learning model for determining the occupancy of an occupancy grid, the machine learning model is trained to predict the occupancy of the occupancy grid at a subsequent time from a history of occupancies; generating a temporal sequence of occupancies of the occupancy grid using measurement data from the environment of the vehicle; evaluating, using the trained machine learning model, the sequence of occupancies that were generated, to predict a total occupancy of the occupancy grid at a current time; comparing the predicted occupancy of the occupancy grid at a current time to the occupancy determined using measurement data, determining a location-dependent measure of a reliability of the occupancy information depending on the comparison; planning a trajectory for the at least partially automated vehicle based on the reliability.