Multi-Sensor Target State Estimation Under Noisy Observations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating the state of a target, such as position, velocity, and orientation, in unmanned driving are inaccurate due to noise in observed data, and fail to efficiently utilize data from multiple sensors, leading to potential safety issues.
Innovation Solution
A method that fuses observation variables from multiple sensors, including image and point cloud acquisition devices, to optimize state variables by minimizing a loss function that includes position, orientation, velocity, and structural constraints, enabling robust and accurate state estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If observation data from multiple sensors is fused to improve state estimation accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent combines observation data from multiple sensors (image sensors, point cloud sensors, radar, etc.) into a unified state estimation framework. The sensor fusion module integrates heterogeneous data sources to improve measurement precision while managing system complexity through a structured optimization approach.
Solution Approach 2:
The patent transforms the complex multi-sensor fusion problem into an optimization problem by defining state variables and loss functions. By changing the parameters to be optimized (position, velocity, orientation, size) and using iterative optimization algorithms, the system achieves accurate state estimation without being overwhelmed by the complexity of direct multi-sensor integration.
2Speed
If real-time state estimation is performed to improve safety response time, then speed is improved, but measurement precision may deteriorate due to noise in observed data
Solution Approach 1:
The patent implements an iterative optimization framework where the loss function provides feedback on the quality of state estimates. The optimization process continuously adjusts state variables based on feedback from multiple sensor observations, enabling real-time estimation while maintaining precision through iterative refinement rather than single-pass processing.
Solution Approach 2:
The patent performs preliminary data processing and feature extraction from sensor inputs before the main optimization step. By pre-processing observation data to extract relevant features and initialize state variables, the system reduces the computational burden during real-time optimization, achieving both speed and precision.
Data Source
AI summary
The present disclosure relates to a method for state estimation of a target comprising: obtaining an observation variable of the target at different moments through a plurality of sensors, wherein at least one observation variable is acquired by each sensor; determining a state variable of the target at different moments based on the observation variable; and optimizing the state variable of the target by minimizing a loss function. The loss function includes at least one of a position loss, an orientation loss, a velocity loss, a size loss, or a structural constraint of the target. The method of the present disclosure may obtain a sufficiently accurate state estimate. In addition, an apparatus, an electronic device, and a medium for state estimation of the target are also provided.