Roles of anticipation and inertia in active elastic sheet models

---

Thesis Type: Postgraduate

Institution Of The Thesis: Middle East Technical University, Graduate School of Natural and Applied Sciences, MAKİNE MÜHENDİSLİĞİ ANABİLİM DALI, Turkey

Approval Date: 2022

Thesis Language: English

Student: MERT DEMİREL

Supervisor: ALİ EMRE TURGUT

Open Archive Collection: AVESIS Open Access Collection

Abstract:

The use of Active Matter (AM) concepts in robotics so as to model collective motion has become very prevalent, with Active Elastic Sheet (AES) models being the most recent ones. In this thesis, anticipation and inertial effects are introduced as the modifying AES parameters prior to multiple allocation studies. It is shown that anticipation behaves similar to damping under certain conditions, and inertial effects introduce a dynamical relaxation equivalent to a proportional velocity controller. Since velocities cannot instantly reach desired values due to the proportional gain, agents respond as if they have inertias. Rotational inertia is also investigated as the interception of instant alignment change apart from mass effects along the velocity vector. Reduced active matter models that utilizes two-particle versions of AES are developed in the pursuit of simpler representations for multi-agent systems. Toy models with both anticipation and inertial effects promise better preliminary studies for swarm robotics applications as they become more and more complex. A multi-agent simulator is built and several scenarios are functionalized in order to assess performance of toy models by comparison. Metrics such as damping ratio and agent separation are studied to discover behavioural similarities between reduced and full models. Such similarities could enable simpler parameter tuning procedures applied on reduced toy models by extrapolating to multi-agent systems more practically. All experiments focus on tuning of anticipation as the control parameter for which order-disorder transition comparisons are carried out in order to understand its gradual effect. The relation between mass included analytical AM models and engineering controllers is justified with reduced versions of AES and phase transition studies under the measure of polarization.