Autonomous Vehicle Acceleration Control for Smooth Object Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional vehicle planning systems often over- or under-estimate object behavior, leading to abrupt acceleration or deceleration, which can result in unpleasant travel experiences and reduced efficiency and passenger comfort.
Innovation Solution
The system determines a planned path for a vehicle and calculates a weighted average acceleration based on the predicted actions of multiple objects in the environment, using sensor data and probabilistic models to assess the likelihood of each object's impact on the vehicle's path.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional planning systems control vehicle deceleration based on detected objects, then collision avoidance is improved, but passenger comfort deteriorates due to abrupt deceleration
Solution Approach 1:
The system dynamically adjusts the deceleration profile by computing a weighted average of deceleration values from multiple prediction systems, where weights are determined by collision probabilities. This creates a smooth, adaptive deceleration curve rather than abrupt fixed thresholds, resolving the contradiction between safety and comfort.
Solution Approach 2:
The system changes the parameter of deceleration from a binary on/off control to a continuous weighted average of multiple prediction outcomes. By varying the deceleration magnitude based on probabilistic assessments, the system achieves both collision avoidance and passenger comfort.
2Device complexity
If conventional systems consider single object predictions, then computational simplicity is improved, but accuracy of behavior estimation deteriorates
Solution Approach 1:
The system merges predictions from multiple independent prediction systems into a unified behavioral estimate. By combining multiple probability assessments through weighted averaging, the system achieves more accurate behavior estimation while managing computational complexity through systematic integration.
Solution Approach 2:
The system uses probabilistic feedback from multiple prediction systems to continuously refine the weighted average deceleration calculation. Each prediction system provides feedback on object behavior likelihood, which is integrated to improve overall estimation accuracy.
3Speed
If conventional systems use threshold-based collision probability, then decision speed is improved, but efficiency of multiple object consideration deteriorates
Solution Approach 1:
The system applies partial consideration of multiple objects by using weighted averaging focused on the most probable collision risks rather than exhaustive analysis of all detected objects. This partial action approach maintains decision speed while improving navigation efficiency through selective probabilistic assessment.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Acceleration determination for controlling a vehicle, such as an autonomous vehicle, is described. In an example, objects in an environment of the vehicle are identified and a probability that each object will impact travel of the vehicle is determined. Individual accelerations for responding to each object may also be determined. Weighting factors for each of the accelerations may also be determined based on the probabilities. A control acceleration may be determined based on the weighting factors and the accelerations.