Estimated-Location Image for Camera-Radar Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to improve the accuracy in recognizing target objects using a combination of cameras and millimeter-wave radars, as they do not effectively integrate sensor data from overlapping sensing ranges to enhance object detection and classification.
Innovation Solution
An information processing apparatus and method that generate an estimated-location image based on sensor data from overlapping sensing ranges, allowing for improved object recognition by integrating data from cameras and millimeter-wave radars, and using machine learning models to enhance detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera and millimeter-wave radar data are combined for target object recognition, then recognition accuracy should improve, but the complexity of data processing and coordinate system alignment increases
Solution Approach 1:
The patent introduces an estimated-location image as an intermediary representation that bridges the camera image coordinate system and millimeter-wave radar coordinate system. This estimated-location image serves as a mediator that contains both camera image information and radar-detected object location information, enabling accurate object recognition without directly processing complex multi-source data fusion. The intermediary approach simplifies the data processing complexity while maintaining high recognition accuracy.
2Reliability
If millimeter-wave radar sensing range is extended to overlap with camera sensing range, then object detection capability improves, but the processing load and computational requirements increase
Solution Approach 1:
The patent extracts only the essential location information from the millimeter-wave radar sensing data and superimposes it onto the camera image coordinate system. Instead of processing the complete radar point cloud or full sensing data, the system extracts key location parameters (distance, angle) and represents them as overlaid markers or highlighted regions in the estimated-location image. This extraction approach maintains reliable object detection capability while significantly reducing computational load and energy consumption.
Solution Approach 2:
The patent applies different processing qualities to different regions of the sensing field. The estimated-location image maintains full camera image quality for visual information while adding localized radar-enhanced information only at specific positions where objects are detected by radar. This local quality enhancement strategy improves object detection capability in radar-detected regions without requiring uniform high-processing-load treatment across the entire sensing range, thereby reducing overall processing load.
3Measurement precision
If coordinate transformation is performed between radar plane and camera-image plane, then data integration accuracy improves, but calculation time and processing complexity increase
Solution Approach 1:
The patent performs preliminary coordinate transformation and estimated-location image generation in advance, before the actual object recognition process. The estimated-location image is pre-computed by transforming radar coordinate data into the camera image coordinate system and superimposing radar-detected object locations onto the corresponding camera image regions. This preliminary action prepares the integrated data structure beforehand, allowing the actual recognition process to proceed faster without time-consuming coordinate transformations, thus reducing processing time while maintaining high data integration accuracy.
Data Source
AI summary
The present technology relates to an information processing apparatus, an information processing method, a program, a mobile-object control apparatus, and a mobile object that make it possible to improve the accuracy in recognizing a target object.An information processing apparatus includes an image processor that generates an estimated-location image on the basis of a sensor image that indicates, in a first coordinate system, a sensing result of a sensor of which a sensing range at least partially overlaps a sensing range of an image sensor, the estimated-location image indicating an estimated location of a target object in a second coordinate system identical to a coordinate system of a captured image obtained by the image sensor; and an object recognition section that performs processing of recognizing the target object on the basis of the captured image and the estimated-location image. The present technology is applicable to, for example, a system used to recognize a target object around a vehicle.