Task-Parameter Nexus: Task-Specific Parameter Learning for Model-Based Control

TL;DR

This paper introduces the Task-Parameter Nexus (TPN), a neural network-based method for online prediction of optimal control parameters in model-based control, enabling robust tracking across diverse tasks.

Contribution

The paper proposes TPN, a novel deep learning approach that predicts control parameters for new tasks, improving adaptability and performance in model-based control systems.

Findings

TPN accurately predicts near-optimal control parameters.

TPN generalizes well to unseen tracking tasks.

Experimental validation on quadrotor shows robustness.

Abstract

This paper presents the Task-Parameter Nexus (TPN), a learning-based approach for online determination of the (near-)optimal control parameters of model-based controllers (MBCs) for tracking tasks. In TPN, a deep neural network is introduced to predict the control parameters for any given tracking task at runtime, especially when optimal parameters for new tasks are not immediately available. To train this network, we constructed a trajectory bank with various speeds and curvatures that represent different motion characteristics. Then, for each trajectory in the bank, we auto-tune the optimal control parameters offline and use them as the corresponding ground truth. With this dataset, the TPN is trained by supervised learning. We evaluated the TPN on the quadrotor platform. In simulation experiments, it is shown that the TPN can predict near-optimal control parameters for a spectrum of tracking tasks, demonstrating its robust generalization capabilities to unseen tasks.