Smart Nodes Augmenting Autonomous Vehicle Perception
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for autonomous vehicle navigation, particularly at intersections, face challenges due to occlusions from static and dynamic objects, leading to incomplete perception and increased costs for deployment and maintenance.
Innovation Solution
A network of 'smart nodes' with computer vision systems and processors is deployed at intersections, generating augmented perception data that includes location and motion information for moving objects, which is communicated to vehicles via a remote server system, enhancing navigation through intersections by providing data on traffic light states and conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If a network of smart nodes with computer vision systems is deployed at intersections, then perception completeness is improved by overcoming occlusions, but device complexity and deployment cost increase
Solution Approach 1:
The system segments the complex monitoring task by deploying distributed smart nodes at different intersections, where each node independently captures images and generates augmented perception data for its local area. This divides the overall system into manageable independent units that collectively provide comprehensive coverage
Solution Approach 2:
A remote server system acts as an intermediary between smart nodes and autonomous vehicles. The server receives augmented perception data from multiple nodes, processes and fuses this data, then communicates relevant information to vehicles approaching intersections, thereby simplifying the architecture and reducing direct complexity at node level
2Reliability
If smart nodes generate and communicate augmented perception data to all vehicles, then navigation safety is improved, but communication bandwidth and processing load increase
Solution Approach 1:
The system applies local quality by providing augmented perception data specifically to vehicles that need it - namely those approaching intersections where smart nodes are deployed. The remote server determines which vehicles require which data based on their location and imminent path, rather than broadcasting to all vehicles uniformly
Solution Approach 2:
Smart nodes continuously capture images and generate augmented perception data in advance before vehicles arrive at intersections. The remote server pre-processes this data and makes it available to vehicles before they reach the intersection, allowing vehicles to receive relevant information proactively rather than reactively
Data Source
AI summary
A system provides navigational control information to a fleet of vehicles and includes a network of nodes within a geographical area. Each node is located at a different intersection in the geographical area. Each node includes a node vision range and a processor which generates augmented perception data (APD) for each moving object of interest monitored in the node vision range. The system includes a remote server system that includes a database of the APD and receives a query from a vehicle of the fleet for the APD associated with an imminent path of the vehicle. The server system, in response to the query, searches the database for APD associated with the imminent path, and communicates, over a communication network, the resultant APD associated with the imminent path to the vehicle. The APD controls navigation of the vehicle through one or more imminent intersections along the imminent path.