3D Point Cloud Remote Control for Faster Moving Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for detecting the position of a moving object by matching a model stored in memory require a long processing time, necessitating a method to shorten this detection period.
Innovation Solution
A remote control device that acquires three-dimensional point cloud data, estimates the position and orientation of a moving object by matching a template point cloud, and determines a start position for matching near the object to quickly generate control commands for remote control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sequential scanning is used to match the two-dimensional model with projected images, then the matching process is systematic and complete, but the processing time becomes excessively long
Solution Approach 1:
The patent applies preliminary action by using detected position information (from previous frames or initial detection) to pre-determine the start position for template matching. This eliminates the need for sequential scanning from a fixed origin, as the matching process begins directly near the object's current position, significantly reducing processing time while maintaining detection accuracy
Solution Approach 2:
The patent changes the parameter of matching start position from a fixed origin-based position to a dynamically determined position near the detected object. This parameter change allows the system to adapt the matching process to the object's actual location, reducing the search space and processing time while preserving measurement precision
2Measurement precision
If template matching is performed across the entire image space, then the position detection is thorough and accurate, but the processing period increases significantly
Solution Approach 1:
The patent extracts and utilizes detected position information to identify and focus the template matching process only on the relevant region near the object's current position. This extraction of position data allows the system to eliminate unnecessary matching operations in other image areas, thereby improving processing speed without sacrificing detection accuracy
Solution Approach 2:
The patent applies local quality by concentrating the template matching resources and computational effort specifically on the local region where the object is detected to be present. Instead of uniformly processing the entire image, the system directs high-quality matching operations only to the relevant local area, optimizing both speed and accuracy
Data Source
AI summary
A remote control device includes, at least one processer configured to perform processes including, acquiring three-dimensional point cloud data measured using a distance measuring device, estimating at least one of a position and an orientation of a moving object in the three-dimensional point cloud data by matching a template point cloud indicating the moving object with the three-dimensional point cloud data, determining a start position to start matching of the template point cloud with the three-dimensional point cloud data; and generating a control command for remotely controlling the moving object using at least one of the estimated position and the estimated orientation of the moving object and transmit the control command to the moving object.