Self-Location Estimation Device Using Reliability-Based Camera Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional self-location estimation techniques for movable bodies often fail, leading to interruptions in movement trajectories, especially when relying solely on camera data without additional sensors.
Innovation Solution
An estimation device that calculates and compares the reliability of self-location estimation results from different methods, integrating them to maintain continuous tracking of the movable body's position and pose using a camera and processing unit, without requiring additional sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional self-location estimation techniques are used, then the system can estimate location from camera images, but the estimation may fail and cause interruptions in movement trajectory
Solution Approach 1:
The system performs preliminary actions by calculating multiple candidate locations and their reliability scores before final determination. When the current estimation reliability falls below the threshold, the system has already prepared alternative locations from previous reliable estimations, allowing seamless transition without trajectory interruption.
Solution Approach 2:
The system provides beforehand cushioning by maintaining a buffer of previously determined reliable locations. When current estimation becomes unreliable, these pre-prepared locations serve as a cushion to maintain continuous trajectory estimation, preventing gaps or interruptions in the movement path.
2Device complexity
If only camera data is used for self-location estimation, then no additional sensors are required, but the estimation reliability decreases and may fail
Solution Approach 1:
The system segments the location estimation task into multiple candidate estimations from different methods or time points, each evaluated independently for reliability. This segmentation allows the system to select the most reliable estimation without requiring additional sensors, maintaining simplicity while improving reliability through diversified estimation approaches.
Solution Approach 2:
The system changes parameters by adjusting the reliability threshold and selecting from multiple candidate locations based on their calculated reliability scores. This parameter-based selection allows the system to maintain high estimation reliability using only camera data, avoiding the need for additional sensors while compensating for the limitations of single-sensor estimation.
3Reliability
If the system switches to alternative estimation methods when reliability is low, then continuous tracking is maintained, but the determination process becomes more complex
Solution Approach 1:
The system implements feedback by continuously monitoring the reliability of current location estimation and automatically switching to alternative candidate locations when the reliability falls below the threshold. This feedback mechanism maintains continuous tracking reliability through a standardized switching process, preventing the determination process from becoming overly complex despite multiple estimation methods.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
According to an arrangement, an estimation device (20) includes an acquisition unit (21), a first estimation unit (22), a first calculation unit (23), a second estimation unit (24), a second calculation unit (25), and a determination unit (26). The acquisition unit (21) acquires an image captured by a camera. The first estimation unit (22) estimates first estimation information from the image. The first calculation unit (23) calculates first reliability of the first estimation information. The second estimation unit(24) estimates second estimation information. The second calculation unit (25) calculates second reliability of the second estimation information. The determination unit (26) decides at least the position or the pose of the camera from the first and second estimation information, based on the first and second reliability, and outputs at least the decided position or pose as a determination result. The second estimation unit (24) estimates the second estimation information based on a past determination result by the determination unit (26).