Cooperative Transport of Unknown Objects by Distributed Effort Regulation in Multi-agent Systems
Authors: Tamzidul Mina, Shyam S. Kannan, Wonse Jo, Shaocheng Luo, Galen B. King, Byung-Cheol Min
Organizations: Sandia National Laboratories, Albuquerque, NM 87123. · This research in its entirety was conducted while previously affiliated with the SMART Lab, Department of Computer and Information Technology, Purdue University, and the School of Mechanical Engineering, Purdue University, West Lafayette, IN 47907, USA. · The Boeing Company, Bengaluru, India. · Department of Information and Telecommunication, Incheon National University, South Korea. · Dept. of Electrical and Computer Engineering, Duke University. · School of Mechanical Engineering, Purdue University, West Lafayette, IN 47907, USA · Department of Computer Science and Department of Intelligent Systems Engineering, Indiana University, Bloomington, IN, USA.
A distributed multi-robot object transportation strategy is proposed for objects of unknown shape, size, geometry, mass and mass distribution without any prior knowledge about the object or number of robots involved in the transportation process. Robots follow the boundary of the object and dynamically determine viable pushing points. A novel object-model-free heading-error-driven pushing effort regulation method is proposed to promote goal-directed object motion; robots regulate their pushing effort using individual and collective heading errors. Robots dynamically relocate around the object boundary if a pushing point is lost or the required pushing effort exceeds its capability to correct its heading. The proposed method is validated using simple and complex shapes with convex and non-convex boundary regions and dynamically changing weight distribution of objects. Detailed analysis on the effects of robot distribution and initial object pose on the transportation performance in terms of required time and path efficiency is also presented. Scalability analysis revealed that object transportation time is dependent on the number of pushing surfaces available which in turn depends on the object size, shape, and set inter-robot safety distance. The proposed system yielded highly efficient object paths with consistent elapsed transport times over several trials robust to external disturbances such as non-uniform friction between contacting surfaces and robot wheel slip.
Figures & tables
Fig. 1 : Conceptual illustration of the proposed multi-robot arbitrary object transportation approach.
Fig. 2 : Finite state machine controller with initial state S1 .
Fig. 3 : Conceptual image of the proposed multi-robot object transportation process for objects with unknown geometric centroid and center of mass (heavy and light ends illustrated by darker and lighter shade).
Fig. 4 : Pushing effort ρ5 model for robots in state S5 based on θi,g and \sign(θnet) for ϕ=30o , α=1 following Eq. 6 .
Fig. 5 : Experiment A.1: Transporting a 0.04 kg disk with diameter 0.83 m and uniform weight distribution using 2 robots. With only one potential pushing location, the object was transported by one robot as expected.
Fig. 6 : Experiment A.2: Transporting a 1.27×0.78 m, 3.5 kg rectangular box with uniform weight distribution using 2 robots. Due to its shape, two pushing locations are available when a side is perpendicular to the goal direction; at other times, the box must be pushed at the corners. The robots are observed to collaboratively transport the object to the goal.
Fig. 7 : Experiment B.1: Two robots placed to find two pushing locations simultaneously; the rectangular box is initially positioned such that the object center to goal vector is perpendicular to its longest edge. The proposed pushing effort regulation resulted in efficient straight paths to the goal over all trials compared to a constant pushing effort.
Fig. 8 : Experiment B.2: An additional weight of 2.2 kg added on one end of the object during the straight line transportation process as in Experiment B.1, to mimic dynamically changing weight and weight distribution of the object. The proposed pushing effort regulation resulted in efficient straight paths to the goal over all trials compared to a constant pushing effort.
Fig. 9 : Experiment C: Arbitrary shaped object with non-uniform weight distribution at θo=30o transported to goal location by 5 robots regulating their pushing effort and relocating if necessary in response to object behavior.
Fig. 10 : Object of non-uniform weight successfully transported to defined goal location with 10 randomly placed robots around the object, for all trials of initial object orientation θo=0o to 330o at increments of 30o .
Fig. 11 : Object transportation to goal locations extended to waypoint following with acceptable cross-track error. Blue, red and white markers denote starting, end and intermediate path positions.
Fig. 12 : Scalability: The object of non-uniform weight with θo=30o was successfully transported to the defined goal with N=1 , 2 , 5 , 10 , 15 and 20 randomly placed robots around the object over 10 trials each.
More-Than-One Robotics Laboratory, Faculty of Sustainable Design Engineering, University of Prince Edward Island, Charlottetown, Canada · Faculty of Electronics and Telecommunications, VNU University of Engineering and Technology, Hanoi, Vietnam