Experiments on multi-agent capture of a stochastically moving object using modified projective path planning

Abstract

Most of the past research in swarm robotics has considered object capture and transport using a specified and very large number of agents. The objects therein were either stationary or moving deterministically (i.e., along a known path). In most previous efforts, the obstacles were also considered stationary. Here we present a modified projective path planning algorithm and illustrate via laboratory experiments that an object exhibiting stochastic (unplanned) but lowspeed motion can be restrained by a limited number of agents guided in real-time across randomly moving obstacles. Relaxation of certain restrictions in the grasping objective allows for the determination of a minimum number and placement of agents around the perimeter of any generically shaped prismatic object. A closed loop experiment is designed using a single overhead camera that provides the visual feedback and helps determine the instantaneous positions of all entities in the workspace. Control signals are sent to the robots via wireless modules by a central processing unit to navigate and guide them to their respective new positions in the subsequent time-step. Agents continue to receive signals until they restrict the moving object in form closure.

An introductory review of swarm technology for spacecraft on‐orbit servicing

“…This is F I G U R E 3 Nanosatellites reclaiming a failed satellite. 40 F I G U R E 4 Example of caging or form closure of arbitrary two-dimensional shapes (reproduced from Shinde et al 54 ).…”

“…However, the literature on capture problems in this field F I G U R E 5 Sequence from an experiment on the capture of an arbitrary two-dimensional object using multiple ground mobile robots: (A) start of the experiment; (B-F) target and obstacles move and encapsulation of target by robots; (G-J) capture of the target. 54 mostly focuses on known objects. The capture of unknown objects by MAS still requires further insights and development.…”

“…Example of caging or form closure of arbitrary two‐dimensional shapes (reproduced from Shinde et al 54 ). …”

“…On the other hand, studies that consider the approach to grasp or cage are more interesting in the context of this research since the capture of target objects in space is what constitutes the gap in the literature, while a few studies have already focused on attitude takeover and detumbling of space objects using multiple agents as previously stated, starting with Shinde et al 54 and Sharma et al, 60 who, in a series of research papers, proposed a method for capturing 2D polygonal objects with multiple robots by planning paths on the projected contour of the object shape. Although some satisfactory results were achieved experimentally, the object's shape in this work was known, and capture points were predefined beforehand.…”

Design and implementation of a millirobot for swarm studies –mROBerTO

“…8 They have been suggested for a variety of application areas, such as dynamic wireless sensor networks (WSNs) with mobile nodes, 9,10 micro-manufacturing with small scale robots 11 , and minimally invasive medical treatments. 12,13 Future applications also include urban [14][15][16][17][18] and wilderness [19][20][21][22][23] search and rescue, and surveillance [24][25][26][27][28][29] .…”