Planar Robot Dead-Reckoning with Sensor Fusion and Kalman Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing navigation systems in planar robots, such as robot vacuum cleaners, suffer from measurement errors and inaccuracies due to wheel slips and low-quality optical flow sensor readings, leading to poor navigation accuracy on heterogeneous surfaces.
Innovation Solution
Fusing velocity measurements from multiple sensors, including optical flow and wheel encoders, with inertial measurement unit data, and applying Kalman filtering to compute a robust and accurate dead-reckoning position estimate, while rejecting faulty sensor readings and weighting measurements based on their accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If wheel encoders and optical flow sensors are used to measure robot speed, then navigation coverage and exploration capability are improved, but measurement precision deteriorates due to wheel slips and sensor failures on heterogeneous surfaces
Solution Approach 1:
The patent combines multiple speed measurement sensors (wheel encoders and optical flow sensors) with an inertial measurement unit (IMU) to create a fused velocity estimate. This merging of sensors compensates for individual sensor failures - when wheel slips occur or optical flow quality degrades, the IMU data provides reliable alternative measurements, maintaining overall measurement precision while preserving navigation productivity
Solution Approach 2:
The system implements feedback by continuously monitoring the quality and consistency of measurements from each sensor source. The Kalman filter uses feedback from sensor measurements to dynamically adjust velocity estimates, rejecting inconsistent readings and weighting reliable measurements higher, thereby maintaining measurement precision across varying surface conditions
2Measurement precision
If multiple sensors are fused to improve speed estimation accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent introduces an intermediary Kalman filter that mediates between multiple sensor inputs and the final velocity estimate. This intermediary processing layer systematically combines measurements from wheel encoders, optical flow sensors, and IMU while filtering out inconsistencies, achieving high measurement precision without requiring complex custom integration logic
Solution Approach 2:
The inertial measurement unit (IMU) serves multiple functions simultaneously: it provides velocity measurements during wheel slips, offers heading information for navigation, and acts as a backup when optical flow sensors fail. This multi-functionality reduces the need for additional specialized sensors, managing device complexity while maintaining measurement precision
3Extent of automation
If dead-reckoning is used to compute robot trajectory, then navigation autonomy is improved, but position estimation accuracy deteriorates due to error accumulation from inconsistent sensor measurements
Solution Approach 1:
The system performs preliminary action by fusing sensor measurements and computing accurate velocity estimates before integrating them into position calculations. By preparing reliable velocity data through sensor fusion and Kalman filtering in advance, the system prevents error accumulation during the dead-reckoning integration process, maintaining position estimation accuracy while preserving navigation autonomy
Data Source
AI summary
Systems and methods provide for estimating a trajectory of a robot by fusing a plurality of robot velocity measurements from a plurality of robot sensors located within a robot to generate a fused robot velocity based on the accuracy of those robot velocity measurements and applying Kalman filtering to the fused robot velocity to compute a current robot location.