Modular Robot Vision-Based Localization to Reduce Trajectory Drift
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autonomous robots face challenges in maintaining accurate navigation and localization, especially when performing tasks that require precise positioning and trajectory maintenance, due to drift in intended trajectories and loss of positional knowledge.
Innovation Solution
A modular robot equipped with a processor that captures and combines images from different positions and orientations using exteroceptive sensors, generates a digital spatial representation of the environment, and estimates a corrected position using a statistical ensemble of simulated positions to address positional drift and maintain accurate navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the robot uses traditional navigation methods with fixed trajectory planning, then the navigation path is simple to compute, but the robot loses positional knowledge due to trajectory drift over time
Solution Approach 1:
The robot continuously captures images, detects overlapping areas between consecutive images, and uses feature matching to compute its actual position. This feedback loop corrects trajectory drift by comparing expected position with actual position derived from visual features, maintaining positional accuracy over time
Solution Approach 2:
The patent replaces traditional mechanical odometry and trajectory-based navigation with a vision-based feedback system. Instead of relying on wheel encoders and pre-planned paths, the system uses image processing and feature matching to determine position, substituting mechanical measurement with optical sensing and computational analysis
2Measurement precision
If the robot captures and processes multiple images to determine position, then positional accuracy is improved, but the processing time and computational load increase
Solution Approach 1:
The system continuously captures and pre-processes images as the robot moves, preparing image data in advance. By maintaining a buffer of recent images and pre-computing feature descriptors, the system reduces real-time processing requirements when position correction is needed
Solution Approach 2:
The robot processes only the necessary subset of images - specifically comparing current image with previous images to find overlapping areas. Rather than processing all captured images, the system focuses computation on relevant image pairs that provide positional information, reducing overall computational load
3Adaptability or versatility
If the robot uses a modular design with interchangeable components, then adaptability to different tasks is improved, but the device complexity increases
Solution Approach 1:
The robot employs a standardized modular architecture where components such as sensors, processors, and effectors can be interchangeably configured for different tasks. The universal base platform with standardized interfaces allows the same robot to perform diverse functions by simply changing task-specific modules, achieving multi-functionality without proportionally increasing overall system complexity
Data Source
AI summary
Provided is a robot including: a chassis; wheels; electric motors; a network card; sensors; a processor; and a tangible, non-transitory, machine readable medium storing instructions that when executed by the processor effectuates operations including: capturing, with at least one exteroceptive sensor, a first image and a second image; determining, with the processor, an overlapping area of the first image and the second image by comparing the raw pixel intensity values of the first image to the raw pixel intensity values of the second image; combining, with the processor, the first image and the second image at the overlapping area to generate a digital spatial representation of the environment; and estimating, with the processor using a statistical ensemble of simulated positions of the robot, a corrected position of the robot to replace a last known position of the robot within the digital spatial representation of the environment.