Linked multi-component mobile robots: Modeling, simulation and control

Quasi-stationary state transportation of a hose with quadrotors

“…Previous works on hose transportation by a team of robots [15] were based on a Generalized Dynamic Splines (GEDS) model of the hose. GEDS are an extension of the Cosserat rods models to capture the dynamics of a DLO.…”

“…Previous works dealt with the control of a team of mobile robots moving in a 2D surface performing the transportation of a hose [15], and control was achieved by the means of Reinforcement Learning [16][17][18][19][20][21][22][23].…”

Cooperative Multi-Agent Reinforcement Learning for Multi-Component Robotic Systems: guidelines for future research

“…They aim to decide the (optimal) joint-action of all the agents to be performed at each time instant. These approaches can be useful for linked systems, such as the Linked MCRS discussed in [23]. Distributed Rewards and Distributed Value Functions were studied in [58] as a way to estimate and control cooperation between neighbors.…”

“…Examples of such tasks are cooperative mapping of an environment [2], establishing dynamic communication links, and driving a hose to a goal [23,24]. Among the desired properties of an MCRS control algorithm, the most salient are:…”

Reinforcement Learning endowed with safe veto policies to learn the control of Linked-Multicomponent Robotic Systems

“…Geometrically Exact Dynamic Splines (GEDS) [38] provide accurate simulation models of hose dynamics [13,12]. The GEDS model of a one-dimensional non-rigid element consists of splines defined by a set of control points and an additional parameter accounting for the hose twist at each point.…”