Florian Kennel-Maushart

PhD candidate

I am a PhD student at the Computational Robotics Lab with Prof. Stelian Coros at ETHZ since June 2020.

I received my Bachelor's degree in Microtechnical Engineering from EPFL in 2013. From 2013-2014, I worked in the research group of Dr. Roderick Lim at the Biozentrum of the University of Basel. After an exchange master thesis at the University of Pennsylvania in the Vijay Kumar Lab, I got my Master's Degree in Microtechnical Engineering with a specialization in Robotics and Automation from EPFL in 2017.
From 2017 to late 2019 I worked as the CTO of now-defunct AR start-up and ETH spin-off RosieReality.

My research interests are centered on Human Robot Interaction, especially for multi-robot systems with applications to construction and installation tasks. I like to work with VR/AR/MR interfaces to facilitate the communication between humans and robots. At the core of my research lies the question, how to use both human intuition and robotic precision and strength to improve human-robot collaboration and make robotic systems intuitive and easy to use.




Multi-Arm Payload Manipulation via Mixed Reality

Florian Kennel-Maushart, Roi Poranne, Stelian Coros

IEEE International Conference on Robotics and Automation (ICRA) - May 2022



Many tasks can be performed better by a multi-robot system than by a single robot. Carrying a large payload is one such task, which is often required on construction sites and in warehouses. Working together, multiple robots can carry these large payloads, while providing more speed and stability and while costing less than a single, specialized robot. Ideally, an operator would only need to specify where to take the payload from, and where to place it, e.g. for installation or storage. However, real environments are generally complex and dynamic, and might require constant attention from the operator. Additionally, directly controlling multiple arms and keeping an overview over their individual and collective motion places a significant cognitive burden on the operator. For example, these systems are sensitive to singular configurations, which can result in dangerous, fast motions. To reduce mental strain for the operator and increase security, it is necessary that, while carrying out the task, the multi-robot system automatically avoids these configurations. We present a mixed reality control interface that allows the operator to specify the payload target poses in real-time, while effectively keeping the system away from bad configurations. To this end, we solve the inverse kinematics problem for each arm individually and leverage available degrees of freedom to optimize for a secondary objective. Using the manipulability index as a secondary objective in particular, allows us to significantly improve the tracking and singularity avoidance capabilities of our multi-robot system in comparison to the unoptimized scenario. We test our approach on different setups and over different input trajectories, and show that the method works well when deployed on to a two-armed ABB YuMi robot.


Manipulability Optimization for Multi-Arm Teleoperation

Florian Kennel-Maushart, Roi Poranne, Stelian Coros

IEEE International Conference on Robotics and Automation (ICRA) - June 2021



Teleoperation provides a way for human operators to guide robots in situations where full autonomy is challenging or where direct human intervention is required. It can also be an important tool to teach robots in order to achieve autonomous behaviour later on. The increased availability of collaborative robot arms and Virtual Reality (VR) devices, provides ample opportunity for development of novel teleoperation methods. Since robot arms are often kinematically different from human arms, mapping human motions to a robot in real-time is not trivial. Additionally, a human operator might steer the robot arm toward singularities or its workspace limits, which can lead to undesirable behaviour. This is further accentuated for the orchestration of multiple robots. In this paper, we present a VR interface targeted to multi-arm payload manipulation, which can closely match real-time input motion. Allowing the user to manipulate the payload rather than mapping their motions to individual arms we are able to simultaneously guide multiple collaborative arms. By releasing a single rotational degree of freedom, and by using a local optimization method, we can improve each arm’s manipulability index, which in turn lets us avoid kinematic singularities and workspace limitations. We apply our approach to predefined trajectories as well as real-time teleoperation on different robot arms and compare performance in terms of end-effector position error and relevant joint motion metrics.


Intrusion detection for stochastic task allocation in robot swarms

Florian Maushart, Amanda Prorok, M. Ani Hsieh, Vijay Kumar

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) - September 2017



We present a novel framework for integrity analysis of swarm robotic systems using the symmetric Kullback-Leibler Divergence. The objective is to understand a robot swarm's vulnerability to malicious intrusion and to develop the necessary computational tools that would detect the presence of malicious agents within the swarm. Using ensemble approaches for modeling and analyzing stochastic task allocation, we analyze the performance of the proposed strategy subject to different system parameters, and show how different design choices can facilitate early intrusion detection. We further evaluate the performance of our method in realistic scenarios through stochastic simulations for different team sizes. The main contribution is an analysis framework whose output can be used to avoid system-inherent design flaws and to decrease the damage that can be inflicted by an undetected attacker.






Intrusion Detection for Stochastic Task Allocation in Robot Swarms

Master Thesis - EPFL / DISAL & UPenn / Kumar Lab - April 2017



Swarm Robotic Systems are recently becoming a reality, after they had mostly been used in theoretical frameworks and simulations due to the high production cost of real robots in the past. Networks of autonomous taxis, quadcopters for autonomous aerial inspection of farms and industrial structures or large scale mobile sensor networks are only a few examples where Swarm Robotic Systems theory might be applied in the near future. While this seems very promising, it is important to understand how well we will be able to control and supervise the behavior of each individual robot when the swarm becomes very large. If the swarm can be disturbed by maliciously changing the behavior of some agents and we are not able to detect this intrusion early enough, the consequences could potentially be detrimental to the performance and security of the system. In this thesis, we therefore present a novel framework for the integrity analysis of Swarm Robotic Systems, using an approximation of the symmetric Kullback-Leibler Divergence. The objective is to understand a robot swarm’s vulnerability to malicious intrusion and to develop the necessary computational tools that would detect the presence of malicious agents within the swarm.





Head Teaching Assistant - Informatik I - ETHZ - WS 2022/23

Teaching Assistant - Informatik II - ETHZ - FS 2022

Teaching Assistant - Visual Computing - ETHZ - WS 2020/21

Teaching Assistant - Distributed Intelligent Systems Lab - EPFL - WS 2015/16







Anwendungen von Mixed- und Virtual-Reality in der Echtzeit-Visualisierung und Steuerung von Multi-Roboter Systemen - 23.06.2022 - Unity Industrieforum Deutschland, Reutlingen

Multi-Arm Payload Manipulation via Mixed Reality - 23.-27.05.2021 - ICRA 2022, Philadelphia

Collaborative Human-Robot Motion Planning with Mixed Reality - 30.03.2022 - Applied ML Days, Microsoft JRC Track Day 2, EPFL, Lausanne

Manipulability Optimization for Multi-Arm Teleoperation - 01.06.2021 - ICRA 2021 (virtual), Xi'an

Digitale Transformation - IoT und Augmented Reality with Thomas Siegrist (Swisscom) - 16.07.2018 - KIWANIS Club Wasseramt, Solothurn

Intrusion detection for stochastic task allocation in robot swarms - 26.09.2017 - IROS 2017, Vancouver


Social Media


sample sample sample