[Paper Review] Collaborative Object Transportation Using MAVs via Passive Force Control

Andrea Tagliabue, Mina Kamel+3 more|arXiv (Cornell University)|Dec 15, 2016

Distributed Control Multi-Agent Systems13 references4 citations

TL;DR

This paper presents a communication-free, passive force control strategy for collaborative object transportation using two micro aerial vehicles (MAVs), where one MAV acts as a master and the other as a slave. The slave uses an admittance controller and a UKF-based force estimator from visual-inertial state estimates to achieve compliance, enabling robust, compliant transport of a 1.2 m long payload without requiring knowledge of payload shape or grasping points.

ABSTRACT

This paper shows a strategy based on passive force control for collaborative object transportation using Micro Aerial Vehicles (MAVs), focusing on the transportation of a bulky object by two hexacopters. The goal is to develop a robust approach which does not rely on: (a) communication links between the MAVs, (b) the knowledge of the payload shape and (c) the position of grasping point. The proposed approach is based on the master-slave paradigm, in which the slave agent guarantees compliance to the external force applied by the master to the payload via an admittance controller. The external force acting on the slave is estimated using a non-linear estimator based on the Unscented Kalman Filter (UKF) from the information provided by a visual inertial navigation system. Experimental results demonstrate the performance of the force estimator and show the collaborative transportation of a 1.2 m long object.

Motivation & Objective

• Enable robust collaborative transportation of bulky payloads using multiple MAVs without relying on inter-MAV communication.
• Overcome limitations of centralized and distributed control by eliminating dependence on payload shape and grasping point knowledge.
• Develop a scalable, bio-inspired master-slave control architecture for MAVs to enhance transportation capabilities.
• Design a computationally efficient, UKF-based external force estimator using visual-inertial state estimates for real-time compliance.
• Demonstrate human-vehicle interaction and collision detection capability without motion capture systems.

Proposed method

• Adopt a master-slave paradigm where the master controls trajectory and the slave enacts compliance via an admittance controller.
• Implement an admittance controller that modifies the slave's trajectory based on estimated external forces and torques.
• Use a non-linear UKF estimator to estimate external forces and torques from MAV state data (position, velocity, attitude, angular velocity).
• Integrate a visual-inertial navigation system (VI-Sensor) to provide real-time state estimates without motion capture.
• Apply quaternion-based UKF formulation to handle attitude dynamics and ensure computational efficiency.
• Detect collisions by thresholding estimated external forces, triggering a safe trajectory re-planning algorithm in under 20 ms.

Experimental results

Research questions

• RQ1 Can collaborative MAV transportation be achieved without inter-vehicle communication?
• RQ2 Can a passive force control strategy enable compliant interaction with a payload using only onboard sensing?
• RQ3 How accurately can external forces be estimated using a visual-inertial system without motion capture?
• RQ4 Can the system detect collisions and react in real time with minimal latency?
• RQ5 Is the approach scalable to multiple slaves and robust under real-world disturbances?

Key findings

• The force estimator achieved RMS estimation errors of 0.43 N, 0.49 N, and 0.57 N for external forces in the x, y, and z directions, respectively.
• Collision detection occurred in less than 20 ms, with detection at 3.887 s after impact at 3.870 s, demonstrating high responsiveness.
• The system successfully transported a 1.2 m long, 0.37 kg cardboard tube using two hexacopters with compliant control on x and y axes.
• The admittance controller maintained stable altitude tracking with a virtual spring stiffness of 10 N/m on the z-axis.
• A peak external torque of 0.8 Nm was detected at t=22 s, corresponding to a propeller collision with the payload.
• Real-time human interaction via a tethered string demonstrated the system's compliance and responsiveness in human-vehicle interaction.