Off-Road Obstacle Assessment Using Dual Machine-Learned Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous off-road vehicles face challenges in navigating through unknown obstacles in off-road environments, as they lack the judgment and experience of human operators to determine whether obstacles can be safely ignored, leading to potential damage or route modifications.
Innovation Solution
The implementation of machine-learned models, including an image classifier and a second model trained on previous interactions, to identify and assess obstacles, determining whether they can be ignored based on dimensions, collision data, and other factors, allowing the vehicle to adjust its navigation or alert remote operators accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If autonomous vehicles avoid all obstacles to reduce damage, then vehicle safety is improved, but navigation efficiency deteriorates due to unnecessary route modifications
Solution Approach 1:
The system changes the parameter of obstacle assessment from binary avoidance to probabilistic risk evaluation. By analyzing multiple parameters (obstacle dimensions, material composition, vehicle speed, proximity), the system determines whether an obstacle poses actual damage risk, allowing safe navigation through environments without unnecessary route modifications.
2Productivity
If autonomous vehicles ignore obstacles to maintain route efficiency, then navigation efficiency is improved, but vehicle damage risk increases
Solution Approach 1:
The system implements feedback through machine-learned models that continuously assess obstacle risk based on sensor data and historical collision outcomes. The models provide real-time recommendations on whether to ignore or avoid obstacles, enabling efficient navigation while maintaining safety through data-driven decision-making.
3Device complexity
If autonomous vehicles use traditional obstacle avoidance algorithms, then implementation simplicity is improved, but adaptability to unknown obstacles deteriorates
Solution Approach 1:
The system employs self-service through machine-learned models trained on historical collision data that autonomously adapt to new obstacle types. The models continuously improve their ability to handle unknown obstacles by learning from past interactions, eliminating the need for complex manual programming while enhancing adaptability.
4Measurement precision
If autonomous vehicles collect and analyze extensive collision data, then obstacle assessment accuracy is improved, but data processing complexity increases
Solution Approach 1:
The system extracts only the most critical features from extensive collision data (obstacle dimensions, material properties, collision force, vehicle speed) for model training. This selective extraction maintains high assessment accuracy while reducing data processing complexity by focusing on the most influential parameters.
Data Source
AI summary
An autonomous off-road vehicle, upon encountering an obstruction while navigating a route, can apply a first machine-learned model to identify the obstruction. In the event that the first machine-learned model cannot identify the obstruction, the autonomous off-road vehicle can apply a second machine-learned model configured to determine whether or not the obstruction can be ignored, for instance based on dimensions of the obstruction. If the obstruction can be ignored, the autonomous off-road vehicle can continue navigating the route. If the obstruction cannot be ignored, the autonomous off-road vehicle can modify the route, can stop, can flag the obstruction to a remote human operator, can modify an interface of a human operator to display a notification or a video feed from the vehicle, and the like.