Optimization Strategies for Real-Time Control of an Autonomous Melting Probe

TL;DR

This paper introduces an optimization-based method for real-time trajectory planning and control of a melting probe, integrating optimal control, MPC-like algorithms, and MINLP fitting, validated through glacier experiments.

Contribution

It presents a novel integrated approach combining optimal control and MINLP fitting for binary control variables in a melting probe system.

Findings

Successful glacier experiment validation

Enhanced model accuracy through automated parameter identification

Effective real-time control demonstrated

Abstract

We present an optimization-based approach for trajectory planning and control of a maneuverable melting probe with a high number of binary control variables. The dynamics of the system are modeled by a set of ordinary differential equations with a priori knowledge of system parameters of the melting process. The original planning problem is handled as an optimal control problem. Then, optimal control is used for reference trajectory planning as well as in an MPC-like algorithm. Finally, to determine binary control variables, a MINLP fitting approach is presented. The proposed strategy has recently been tested during experiments on the Langenferner glacier. The data obtained is used for model improvement by means of automated parameter identification.