Martin Riedel

I am currently a PhD student in the at the Max-Planck-Institut for Biological Cybernetics.

My research interest generally focuses around the interaction of a human operator with a micro aerial vehicle (MAV). These devices can exhibit varying levels of autonomy, ranging from manual operation to complete unsupervised flight without a human operator present.

In my work, I’m developing human-robot interfaces that enable an operator to control an MAV with varying degrees of autonomy. This results in an easier teleoperation and broadens the possible applications of the MAV.

Additionally, the interaction with the operator should be intuitive for each mode of operation. Therefore I’m systematically evaluating interface designs on usability and effectiveness. Currently I’m working with handheld devices of cellphone and tablet size.

Part of my research also focuses on the operator’s perception of the MAV. For example, presenting the user with varying view angles (e.g. first-person vs third person) and augmenting additional information into the interface.

1. Novel Touch Interface Design for UAV Controls

Part of my research is the design of novel interfaces for micro aerial vehicle (MAV) control. We are mainly evaluating the use of touch-based devices in handheld- and tablet-size. Controlling a single robot with such a device may still be a feasible task, however, operating an increasing number of MAVs poses a challenge towards interface design, as well as Swarm / Formation Control Algorithms.

2. TeleKyb Framework

TeleKyb is a software framework that provides a standardized interface for MAV control algorithm development and testing. It is completely based on the , which mainly takes care of different inter-process communication, as well as of dynamic loading of modules at runtime. Additionally, ROS provides a large library of software related to robot control that can easily be integrated with TeleKyb.

Our system abstracts several aspects of control into distinct parts, that are dynamically made available to the TeleKyb System. They are:

• StateEstimator (provides the estimated State to the TeleKyb System. e.g. SimulatorStateEstimator, ViconStateEstimator ...)
• Behaviors (generate the Trajectory Message. An experiment is defined by a cascade of Behaviors. e.g. Ground -> Liftoff -> Brake -> Hover -> FlyCircle -> Break -> Hover -> Land -> Ground)
• Trajectory Modules (Trajectory Modules are placed behind Behaviors and observe, process or modifiy the Trajectory Message that is produced by the currently active Behavior. This enables the user to dynamically react to certain events. e.g. saturate inputs, avoid obstacles, observe and react to differences in current state and commanded trajectory)

This flexible approach is able to satisfy diverse experimental requirements, without sacrificing the advantages offered by a unified controller.

A poster overview of TeleKyb and other Software used in our lab can be found .

3. MAV Control in Virtual Enviroments

In cooperation with the Group, we are evaluating the use of Virtual Enviroments (VE) for the teleoperation of a (swarm of) MAVs. This enables the remote operator to get into proximity to the commanded robots and enables new possibilities for the development of novel Human-Robot-Interfaces in VEs.

4. Multiple-UAV Formation Control over an Unreliable Network

Our group uses TeleKyb in conjunction with several interface methods (haptic device, handheld-device, VR, ...) to control a (swarm of) robots via an unreliable network. Our research focuses on the design of algorithms that are robust towards a certain degree of package delay and/or loss.