Sparse Scale Estimation for Low-Power Object Pose Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Computing devices face challenges in accurately tracking objects due to perspective ambiguity and resource-intensive scale estimation, particularly in devices with limited resources like head-worn XR devices, leading to excessive computational load and power consumption.
Innovation Solution
Implementing temporally sparse scale estimation techniques that reduce the frequency of new scale estimates by maintaining a time-windowed, weighted average of past triangulation results and dynamically switching between multi-camera and single-camera modes, leveraging temporal sparsity to balance computational load and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If scale estimation is performed frequently to improve tracking accuracy, then measurement precision is improved, but use of energy increases
Solution Approach 1:
The system performs scale estimation periodically based on temporal sparsity criteria rather than continuously. Scale estimation is triggered only when temporal thresholds are met (e.g., when a certain time duration has elapsed since the last estimation or when object motion exceeds a threshold), reducing computational frequency while maintaining tracking accuracy when needed.
Solution Approach 2:
The system dynamically adjusts the frequency and parameters of scale estimation based on object characteristics, motion state, and temporal context. By changing estimation parameters adaptively (such as adjustment frequency, reference object selection, and temporal windows), the system optimizes the balance between tracking precision and energy consumption.
2Measurement precision
If scale estimation is performed frequently to improve tracking accuracy, then measurement precision is improved, but computational load increases
Solution Approach 1:
The system implements periodic scale estimation with temporal sparsity, performing estimations only when necessary based on time intervals and object motion criteria. This reduces the computational frequency of complex scale estimation operations while maintaining adequate tracking accuracy through selective updates.
Solution Approach 2:
The system extracts and utilizes temporal patterns and sparsity characteristics from object motion data to determine when scale estimation is necessary. By analyzing temporal intervals, motion thresholds, and prediction confidence levels, the system extracts key temporal features that guide selective scale estimation, reducing unnecessary computational operations.
3Measurement precision
If scale estimation is performed frequently to improve tracking accuracy, then measurement precision is improved, but operational time decreases
Solution Approach 1:
The system performs scale estimation periodically based on temporal sparsity criteria, extending battery-operated device operational time by reducing the frequency of computationally intensive operations while maintaining adequate tracking accuracy through selective updates triggered by temporal thresholds.
Solution Approach 2:
The system uses its own temporal data and motion prediction capabilities to self-determine when scale estimation is necessary, autonomously adjusting estimation frequency based on observed object characteristics, motion patterns, and prediction confidence without external intervention.
Data Source
AI summary
Examples in the present disclosure relate to temporally sparse scale estimation for object tracking. A computing device detects a current orientation of an object. The computing device determines that a difference between the current orientation and at least one of a plurality of previously detected orientations of the object is less than a threshold value. Each previously detected orientation has a respective scale estimate. In response to determining that the difference is less than the threshold value, the computing device generates an effective scale estimate for the object based on a combination of the respective scale estimates for the plurality of previously detected orientations. Each respective scale estimate contributes to the effective scale estimate according to a respective difference between the current orientation and the previously detected orientation for the respective scale estimate. The computing device tracks a pose of the object based on the effective scale estimate.