Autonomous Vehicle Blind Spot Driving Using Server-Based 3D Occlusion Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately recognizing and responding to blind spots due to variations in sensor types, performance, and installation locations, as well as inaccuracies in precision maps.
Innovation Solution
A method for driving in blind spots using communication with a server, where the autonomous vehicle transmits data about its location, sensor installation, and environment to determine if a region of interest is within the blind spot, and adjusts its speed based on stereoscopic blind spot data received from the server.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If autonomous vehicles use precision maps and onboard sensors to detect blind spots, then they can identify occluding regions, but accuracy varies due to differences in sensor types, performance, and installation locations across vehicles
Solution Approach 1:
A server acts as an intermediary between the autonomous vehicle and the blind spot detection system. The server receives sensor data from the vehicle, performs centralized blind spot analysis using standardized algorithms, and returns detection results. This mediator approach standardizes the detection process across vehicles with different sensor configurations, ensuring consistent accuracy regardless of individual vehicle variations.
Solution Approach 2:
The system creates a virtual copy of the blind spot detection capability on the server, which then provides detection results back to the vehicle. Instead of relying on each vehicle's onboard processing, the server generates standardized detection outputs that replicate accurate blind spot identification across all vehicles, eliminating the need for each vehicle to independently achieve the same detection precision.
2Measurement precision
If autonomous vehicles transmit detailed sensor data and precision map information to determine blind spots, then detection accuracy improves, but data transmission volume increases
Solution Approach 1:
The system extracts only the essential data elements needed for blind spot detection from the vehicle's sensor suite and transmits them to the server. Instead of sending complete sensor datasets, the extraction process identifies and transmits only the critical parameters (such as occluding object positions, sensor angles, and relevant environmental features), significantly reducing transmission volume while preserving detection accuracy.
Solution Approach 2:
The vehicle performs preliminary data processing and filtering before transmission, pre-identifying only the data elements that are relevant to blind spot detection. This preliminary action eliminates unnecessary data from the transmission stream, reducing the quantity of transmitted information while ensuring that all essential detection parameters are included.
3Speed
If autonomous vehicles process blind spot information locally using onboard computing devices, then real-time response is achieved, but computational complexity and power consumption increase
Solution Approach 1:
The server serves as a computational intermediary that handles the complex blind spot analysis algorithms centrally. The vehicle's onboard computing device only needs to communicate with the server and receive pre-processed detection results, significantly reducing the computational burden on the vehicle while maintaining real-time response capabilities through efficient server-vehicle communication.
Data Source
AI summary
A method for driving in a blind spot of a sensor mounted on an autonomous vehicle is provided. The method includes steps of: a computing device of the autonomous vehicle running on a specific road locating the autonomous vehicle from precision map information, sensor information and GPS information, and in response to determining that the autonomous vehicle is expected to encounter a specific event, transmitting vehicle location data, travelling direction data, vehicle structure data and sensor location data and sensor's viewing angle data to the server, to determine whether a region of interest corresponding to the specific event is included in blind spot candidates; receiving blind spot stereoscopic data, computed from the data received from the autonomous vehicle and 3D occlusion environmental data corresponding to occluding static objects in the blind spot candidates, from the server; and controlling movement of the autonomous vehicle based on the blind spot stereoscopic data.