J. Joe Payne,James Zhu,Nathan J. Kong,Aaron M. Johnson
Abstract:In this paper we present Hybrid iterative Linear Quadratic Estimation (HiLQE), an optimization based offline state estimation algorithm for hybrid dynamical systems. We utilize the saltation matrix, a first order approximation of the variational update through an event driven hybrid transition, to calculate gradient information through hybrid events in the backward pass of an iterative linear quadratic optimization over state estimates. This enables accurate computation of the value function approximation at each timestep. Additionally, the forward pass in the iterative algorithm is augmented with hybrid dynamics in the rollout. A reference extension method is used to account for varying impact times when comparing states for the feedback gain in noise calculation. The proposed method is demonstrated on an ASLIP hopper system with position measurements. In comparison to the Salted Kalman Filter (SKF), the algorithm presented here achieves a maximum of 63.55% reduction in estimation error magnitude over all state dimensions near impact events.
What problem does this paper attempt to address?
### Problems the paper attempts to solve
The paper aims to solve the state estimation problem in hybrid power systems, especially near the contact surface, due to the non - smoothness and potential discontinuity of dynamics caused by the intermittent contact between the robot and the environment. Traditional state estimation methods (such as the Kalman filter) assume that the system dynamics are smooth, which is not always true in hybrid power systems. Therefore, these traditional methods may have large estimation errors when dealing with hybrid power systems.
### Specific problems
1. **Non - smoothness of hybrid power systems**: When the robot contacts the ground or other objects, its dynamics will change suddenly, making the state estimation of the system complicated.
2. **Accurate estimation of contact events**: Near the contact event, a small change in state may lead to a huge difference in control effects. For example, when the robot's foot contacts the ground, it can exert a large ground reaction force, but when it is slightly off the ground, it cannot exert any force.
3. **Limitations of existing methods**: Existing online state estimation methods (such as the salting Kalman filter, SKF) perform poorly in handling the state estimation of hybrid power systems, especially near the contact event.
### Solutions
To solve these problems, the author proposes the Hybrid Iterative Linear Quadratic Estimation (HiLQE), which is an optimization - based offline state estimation algorithm. The main features of HiLQE include:
1. **Using saltation matrix**: Calculate the gradient information through the saltation matrix to accurately handle the mixed events in the backward pass.
2. **Mixed dynamics in the forward pass**: In the forward pass, the algorithm considers the mixed dynamics and applies the correct system dynamics at each time step.
3. **Reference expansion method**: To handle the state comparison in different modes, the algorithm uses the reference expansion method to calculate the feedback gain, thus handling the noise estimation more accurately.
### Experimental results
The author verified the effectiveness of HiLQE on two systems:
1. **Bouncing ball system**: In 1,000 trials, HiLQE reduced the maximum error near the contact event by 48.60% and improved the median mean - squared error (MSE) by 30.48%.
2. **ASLIP hopping system**: In 100 trials, HiLQE reduced the maximum error near the contact event by 63.55% and improved the median MSE throughout the running process by 61.96%.
### Conclusion
By explicitly considering the characteristics of hybrid power systems, HiLQE significantly improves the accuracy of state estimation, especially outperforming existing online methods (such as SKF) near the contact event. Future research directions may include extending this method to online state estimation and other types of hybrid power systems.