# Neuromusculoskeletal Control for Simulated Precision Task versus Experimental Data in Trajectory Deviation Analysis

## TL;DR

This paper compares simulated robotic arm control with human experimental data to analyze precision task performance and error behavior.

## Contribution

A novel comparison of Feedback Linearization Control and Sliding Mode Control in a musculoskeletal robotic model against human experimental data.

## Key findings

- Feedback Linearization Control performed better than Sliding Mode Control in simulated precision tasks.
- Simulated model errors increased with amplitude but not with frequency, unlike human trials where errors rose with both.
- Discrepancies were observed between simulated and real experimental data in trajectory deviation analysis.

## Abstract

Control remains a challenge in precision applications in robotics, particularly when combined with execution in small time intervals. This study employed a two-degree-of-freedom (2-DoF) planar robotic arm driven by a detailed human musculoskeletal model for actuation, incorporating nonlinear control techniques to execute a precision task through simulation. Then, we compared these simulations with real experimental data from healthy subjects performing the same task. Our results show that the Feedback Linearization Control (FLC) applied performed satisfactorily within the task execution constraints compared to a robust nonlinear control technique, i.e., Sliding Mode Control (SMC). On the other hand, differences can be observed between the behavior of the simulated model and the real experimental data, where discrepancies in terms of errors were found. The model errors increased with the amplitude and remained unchanged with any increase in the task execution frequency. However, in human trials, the errors increased both with the amplitude and, notably, with a drastic rise in frequency.