MIRMI News

Riddhiman Laha on the dream of a robot

Health, Research, Robotics, Artificial Intelligence |

On the one hand, seeing the capabilities of humans, on the other hand, using the strengths of robots: That is the goal of PhD student Riddhiman Laha at the Munich Institute of Robotics and Machine Intelligence (MIRMI).

Riddhiman, you studied mechanical engineering in India and then did your master's in robotics in the US. How did it come about?

Mechanisms and Theory of Machines, a sub-field of Mechanical engineering, is a kind of precursor to robotics. After all, robots are machines that first of all reliably perform exactly what humans built them for. The next logical step is to teach them a perception that is close to that of humans. In Project X within geriatronics research, MIRMI is now about planning how a robot moves in anthropomorphic settings, how it grips and turns its arms and takes over tasks. The particular challenge is that the environment is constantly changing. People are walking around, doing tasks. The robot must learn not to be distracted by these changes in the environment. At the same time, it has to register when it gets in the way of people - if only for safety reasons.

Why is cooperation between humans and robots worthwhile at all?

Humans are very good at understanding relationships, whereas robots can perform work steps very precisely, repeat them as often as they like without getting tired and plan the movements optimally. Bringing human input and robotic precision together in a human-robot interaction scenario and bridging the gap by exploiting geometric representations in robotics: This is also an important part of my doctoral thesis.

Service robot in a human-in-the-loop concept

What is still lacking at the moment?

A robot's perception of its immediate environment can only be as good as the sensors it can use and the resulting ability to model the environment. He has to recognise all objects in a room and learn scenarios that can arise when working with people. We are now starting to build up a database of scenarios. The aim is to develop service robots in a human-in-the-loop concept. Efforts should also be targeted towards utilizing geometric formulations that capture information in a coherent manner and reduce the gap between modelling and control. Humans and robots should work together in a safely modelled system that is efficient but not too slow. After all, it doesn't make sense if the robot constantly stops because it goes into safety mode.

Where will and should all this lead?

The fundamental question for me is: What would be the intrinsic use case for a person wanting to use a robot in, say, a remote village in India or less developed countries? What is the dream that a robot should one day be able to fulfill? When humans and robots come together, it will be important that the robot improves human performance in a synergistic fashion. It doesn't really do that when it puts plates and cups into a dishwasher, but for example when the robots enable a specially-abled person to grasp again, for example through an intelligent prosthesis. For example, when brain waves or muscle activity are used to control and move limbs such as the hand. This is an incredibly exciting field of research in which a lot will happen in the future. But by far not the only one.

Fast Facts - Riddhiman Laha

-        Riddhiman was involved in research related to the Moroccan tea tablet.

-        Riddhiman is one of the initiators of Talking Robots, an online event in which young researchers present current topics in robotics about twice a month.

-        Recent paper: Integrated Bi-Manual Motion Generation and Control shaped for Probabilistic Movement Primitives, u.a. Riddhiman Laha, Luis F.C. Figueredo und Sami Haddadin; IEEE RAS Humanoids, 2022 - Finalist Best Interactive Paper award.

-        Co-organising a Workshop on Geometric Representations: The Roles of Modern Screw Theory, Lie Algebra and Geometric Algebra in Robotics, IEEE ICRA 2023, Full Day Workshop, London