Towards Differentiable Resampling
This addresses a key bottleneck in differentiable particle filters for end-to-end learning, with potential applications in robotics and state estimation.
The paper tackled the non-differentiability of resampling in particle filters by introducing a learned neural network resampler, which outperformed traditional methods on synthetic data and a simulated robot localization task.
Resampling is a key component of sample-based recursive state estimation in particle filters. Recent work explores differentiable particle filters for end-to-end learning. However, resampling remains a challenge in these works, as it is inherently non-differentiable. We address this challenge by replacing traditional resampling with a learned neural network resampler. We present a novel network architecture, the particle transformer, and train it for particle resampling using a likelihood-based loss function over sets of particles. Incorporated into a differentiable particle filter, our model can be end-to-end optimized jointly with the other particle filter components via gradient descent. Our results show that our learned resampler outperforms traditional resampling techniques on synthetic data and in a simulated robot localization task.