Autonomous Navigation Using Clutter Maps for Collision-Aware Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating cluttered environments due to the high computational overhead and cost of advanced sensing and perception technologies, which are often limited in handling complex environments and require accurate object classification and tracking.
Innovation Solution
The approach involves creating a dynamic 'clutter map' that characterizes the risk of encountering objects not represented in the static map, using minimal sensor data and light computational loads, based on risk density and velocity fields, and clustering techniques, allowing for route planning that avoids high-risk areas without requiring accurate object classification or tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If advanced sensing and perception technologies are used to detect and adapt to moving agents and obstacles, then navigation reliability in cluttered environments is improved, but system cost and device complexity increase significantly
Solution Approach 1:
The patent extracts only the essential information needed for navigation (occupancy density and velocity field) from the complex sensor data, rather than processing complete object models. This selective extraction reduces computational complexity while maintaining navigation reliability by focusing on the most critical parameters for obstacle avoidance.
Solution Approach 2:
The patent uses inexpensive sensor data (e.g., from standard LIDAR or depth cameras) to create temporary occupancy grids that are continuously updated and discarded. Instead of relying on expensive, permanent sensing infrastructure, the system generates and uses transient occupancy representations that are computationally lightweight and cost-effective.
2Measurement precision
If accurate object tracking and classification algorithms are implemented, then measurement precision of object locations and velocities is improved, but computational overhead increases significantly
Solution Approach 1:
The patent merges multiple object tracking tasks into a single occupancy density estimation process. Instead of individually tracking and classifying each object, the system combines all object information into a unified occupancy grid that represents the probability of space occupation. This merging reduces computational energy by eliminating redundant processing while maintaining sufficient precision for navigation decisions.
Solution Approach 2:
The patent applies partial action by estimating only the necessary parameters (occupancy density and velocity field) without performing complete object classification and tracking. This selective processing achieves sufficient measurement precision for navigation purposes while significantly reducing computational overhead by avoiding excessive processing of unnecessary details.
3Reliability
If the vehicle plans routes avoiding high-risk locations based on clutter map, then collision risk is reduced, but navigation time increases due to additional risk assessment computations
Solution Approach 1:
The patent performs preliminary action by pre-computing the clutter map and risk density fields before actual route planning. The occupancy density and velocity fields are estimated in advance, allowing the route planner to quickly query preprocessed risk information rather than performing complex real-time simulations. This preliminary preparation reduces navigation time while maintaining collision avoidance reliability.
Solution Approach 2:
The patent introduces an intermediary clutter map that mediates between raw sensor data and route planning decisions. This intermediate representation (occupancy density field) acts as a simplified model that captures essential risk information without requiring full physical simulations. The intermediary structure enables fast risk assessment during route planning, reducing navigation time while preserving collision avoidance capability.
Data Source
AI summary
An approach to autonomous navigation of a vehicle augments a static map of an environment with a clutter map characterizing a risk of encountering an object that is not represented in the static map of the environment. For example, the clutter map may be based on locations and velocities of those objects, and route planning may avoid planning a path through locations that have a high risk of occupancy, and therefore potential delay or collision.