On the Weighting of Control Signals in a Multi-robot System: A Formation-Based Analysis
- PDF / 2,286,946 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 89 Downloads / 170 Views
On the Weighting of Control Signals in a Multi-robot System: A Formation-Based Analysis Valentim Ernandes-Neto1 · Gabriel V. Pacheco2 · Alexandre S. Brandão2 Received: 11 January 2020 / Revised: 23 May 2020 / Accepted: 27 May 2020 © Brazilian Society for Automatics–SBA 2020
Abstract This work presents a strategy to guide an n-robot convoy in a rigid virtual structure, split into n −2 triangular formations. Each set of three robots represents one formation, and transformation functions relate the robots’ position to the formation’s pose and shape, and vice versa. Each triangular formation has its own constructive sequence (clockwise or counterclockwise), which depends on the robots’ position, and is crucial for the control navigation. As first contribution, we propose a strategy based on cross-product concept to identify automatically the current and the desired triangle sequence. Furthermore, once knowing each robot can belong up to three formations, our second contribution is a control signals weighting, to improve simultaneous convergence of the whole formation. Finally, we present real experiments with four Pioneer-3DX robots performing a trajectory tracking task to validate our proposal. Keywords Multi-agent systems · Mobile robot · Cooperative control · Virtual structure formation
1 Introduction Cooperative robot systems have increasingly attracted researchers worldwide. The main reason for the growth of interest is that a group of mobile robots working in cooperation is able to perform certain tasks better than a unique specialized robot, like in military and civilian fields (Bayındır 2016; Couceiro et al. 2014). Besides that, a group can possibly overcome the failure of a robot and complete the desired task. Therefore, a multi-agent system can guarantee robustness by its scalability. Some of these tasks in which groups of robots can perform effectively are surveillance of wide areas (Saska et al. 2016; Acevedo et al. 2013; Wallar et al. The authors thank CNPq, CAPES and FAPEMIG.
B
Alexandre S. Brandão [email protected] Valentim Ernandes-Neto [email protected] Gabriel V. Pacheco [email protected]
1
Department of Informatics, Federal University of Viçosa, Viçosa, Brazil
2
Nucleus of Specialization in Robotics, Department of Electrical Engineering, Federal University of Viçosa, Viçosa, Brazil
2015), inspection of hostile environments (Quater et al. 2014; Deng et al. 2014), cargo transport (Tagliabue et al. 2017; Teixeira et al. 2016), search and rescue operations (Cui et al. 2015; Beck et al. 2016; Baranzadeh and Savkin 2017), and so on. The formation control of multiple agents for cooperative tasks is generally addressed in three different approaches: leader–follower, virtual structure (V-S) and consensus-based strategies. Each of these strategies has its own advantages and drawbacks. Leader–follower approach is very simple to be implemented, but lacks robustness, since all followers depend on the main leader movement. Thus, if the leader fails, all the formation is compromi
Data Loading...
