01
Human–Robot Interaction
Interaction methods that let people and mobile robots collaborate naturally — speech, gesture, pointing, and following, combined into one coherent interface.
- Multimodal interfaces
- Verbal & non-verbal
- Trust & usability
Associate ResearcherRobotics & AI
Building intelligencefor the physical world.
I build robots that people can actually talk to, point at, and work beside — from multimodal human–robot interaction in the laboratory to autonomous navigation in the field and the perception systems underneath.
- Peer-reviewed publications
- 2
- Research systems built
- 9
- Years in robotics research
- 5
Currently at CELISCA, University of Rostock
SYSTEM // AAK-01ONLINE
01
02
03
ROBOTICSAIRESEARCH
PERCEPTION84.7
ACTIONREADY
SenseReasonAct
Human–Robot InteractionMultimodal PerceptionAutonomous NavigationModel Predictive ControlSensor FusionApplied Deep Learning
01 / ABOUT
Research with a systems perspective.
A professional portfolio for research peers, collaborators, and future teams — connecting peer-reviewed research to the systems, experiments, and engineering behind it.
PORTRAIT // 01My work sits where research questions meet real-world constraints.
I am an Associate Researcher at the University of Rostock's Center for Life Science Automation, where I develop the human–robot interaction layer for mobile robots serving laboratory stations across a multi-storey building. Before that I worked on agricultural robot navigation, automotive perception, and optimal control — each time carrying a method from a paper through to something that runs on real hardware.
ROLEAssociate Researcher
DOMAINRobotics & AI
APPROACHResearch to prototype
LANGUAGESEnglish · German · Hindi · Urdu
02 / RESEARCH LENS
From algorithms to embodied systems.
Four threads that run through the publication record and the systems below.
02
Gesture & Activity Recognition
Skeleton and keypoint sequences, temporal models, and attention mechanisms that turn human motion into reliable robot commands.
- Pose estimation
- LSTM + attention
- Dataset design
03
Mobile Robot Autonomy
Planning, control, and state estimation for ground robots — from nonlinear MPC on a lab UGV to Nav2 stacks running in real fields.
- Nav2 & planning
- NMPC
- EKF & visual SLAM
04
Perception & Sensor Systems
Getting trustworthy data out of real hardware: camera degradation modelling, image restoration, and hardware-triggered multi-sensor capture.
- PSF estimation
- GAN deblurring
- Time synchronisation
03 / SELECTED WORK
Nine systems, one throughline.
Every project here was built and evaluated on real hardware or real data — spanning human–robot interaction, mobile robot autonomy, perception, and the research foundations underneath.
Showing 9 of 9 projects.
PROJECT // 01VERBAL + NON-VERBAL HRIMOLAR: Multimodal HRI for a Mobile Laboratory Robot
CELISCA, University of Rostock · ADAM project
MOLAR — the Mobile Laboratory Robot — is an ASTI ProBot L platform that moves samples between stations across a multi-storey laboratory building. I build its interaction layer: verbal and non-verbal channels combined into one interface, so an operator can speak to the robot, gesture at it, or simply be followed by it.
- Voice-ID authentication before any command is accepted
- Speech control for transport creation, station pick-and-place, capabilities, and status
- Gesture control for manual driving: start, turn, reverse, stop, and pointing
- Person-following behaviour with hand-off back to verbal interaction
PROJECT // 0211 GESTURES / 3,300 SAMPLESGESTRO: Skeleton-Based Gesture Teleoperation
Otto von Guericke University Magdeburg · TIAGo platform
A dynamic hand-gesture recognition system for teleoperating a ROS-based TIAGo robot, with a Siamese face-recognition module that only accepts commands from a verified operator. Started as my master's thesis and grew into a peer-reviewed system with an 11-gesture, 3,300-sample public dataset.
- MediaPipe holistic skeleton features with custom feature engineering
- LSTM with a self-attention mechanism for temporal gesture classification
- Face recognition as a security gate on the command channel
- 94.7% accuracy in the thesis evaluation; 98% in the published GESTRO system
PROJECT // 03HUMAN-FOLLOWING VEHICLE
Remote Control for an Autonomous Cargo Cycle
Otto von Guericke University Magdeburg · AuRaHBW with AuRaSys
A Bluetooth remote-control system for a human-following autonomous cargo cycle. The controller is built around an explicit state machine so the vehicle transitions safely between manual, assisted, and autonomous following modes.
- State machine governing manual, assisted, and follow modes
- Bluetooth link between the handheld controller and the vehicle
- ROS 2 and C++ implementation on the cargo cycle platform
Code not publicly released
PROJECT // 04NAVIGATION / SENSOR FUSIONAutonomous Robot for Organic Farming
Otto von Guericke University Magdeburg with aimess GmbH
An agricultural mobile robot that automates soil-sample collection to build detailed fertiliser profiles for a field. I worked across the kinematic model, the local planner, the control design, and the state-estimation stack, then took it into field testing.
- Kinematic model covering the platform's different driving modes
- Local planner and Nav2 tuning for row-following and headland turns
- EKF fusing IMU, wheel odometry, GPS, and visual SLAM for global and local localisation
- Gazebo simulation with custom Blender assets, then iterative field trials
Code not publicly released
PROJECT // 05SHARPNESS / CAMERA RELIABILITYAutomotive Camera Health Monitoring
Technische Hochschule Ingolstadt
A computer-vision pipeline that estimates the point-spread function directly from in-field automotive camera images, uses it to restore sharpness, and tracks how the PSF drifts over time as an early warning that a camera is ageing.
- Blur-kernel (PSF) estimation from real in-field imagery
- Image-quality metrics, feature extraction, and template matching
- GAN-based deblurring to recover usable image quality
- Camera-ageing prediction from changes in the estimated PSF
Code not publicly released
PROJECT // 06SENSOR SYNCHRONISATIONHardware Triggering & Multi-Sensor Time Sync
Technische Hochschule Ingolstadt · research sensor vehicle
The acquisition side of the same research vehicle: a hardware triggering and time-synchronisation system that keeps LiDAR, cameras, GPS, and radar on a shared clock, so every recorded frame can be trusted for mapping and navigation work.
- Arduino clock-pulse generator driven from GPS, running at selectable frequencies
- ROS 2 driver nodes built on the vendor SDKs for each sensor
- Synchronised capture verified live in RViz across cameras and point clouds
Code not publicly released
PROJECT // 07OPTIMAL CONTROL / UGVNonlinear MPC Path Following on the Hamster UGV
Otto von Guericke University Magdeburg
Path planning and optimisation with the ACADO toolkit in MATLAB, and a nonlinear model-predictive controller that follows the resulting trajectory on a real Hamster ground vehicle tracked by an OptiTrack motion-capture system.
- Path planning, obstacle avoidance, and trajectory optimisation via the ACADO solver
- NMPC built on a kinematic bicycle model with rigid-body dynamics
- ROS bringup for hardware execution on the physical UGV
- OptiTrack motion capture for indoor localisation and ground truth
PROJECT // 08SAFETY & ASSURANCE
Safety Analysis & Assessment of Autonomous Vehicles
Otto von Guericke University Magdeburg · scientific project
A structured review of how automated driving systems are argued to be safe: the standards landscape, the reference architectures behind it, and where the open problems still sit.
- ISO 26262 functional safety and ISO 21448 (SOTIF)
- SAE J3016 automation levels and SAE J3061 cybersecurity guidance
- UL 4600 safety-case methodology for autonomous products
- Autonomous-vehicle architectures, open challenges, and state-of-the-art responses
PROJECT // 09NETWORK FOUNDATIONS
Optical Transport Networks & SDN
G D Goenka University · B.Tech final-year project
My undergraduate final project: a technical report on optical transport networking and software-defined networking, written alongside field placements with Alcatel-Lucent Enterprise and NCR Corporation on live carrier infrastructure.
- OTN and SDH transport hierarchies
- DWDM, FTTH, and GPON access and backbone technologies
- Software-defined networking as a control-plane shift
Code not publicly released
04 / RESEARCH IN MOTION
Short demonstrations, real systems.
Privacy-reviewed extracts from the research record. Playback starts only when you choose it — the two MOLAR clips carry the spoken dialogue, the rest are silent.
MOLAR — Speech-Driven Laboratory Transport
The robot is addressed by voice: it reports capabilities and status, accepts direct movement commands, and then switches into person-following mode.
MOLAR — Gesture Control & Pointing
Non-verbal control of the same platform: a thumbs-up hands over manual control, then turn, reverse, stop, and pointing gestures drive the robot directly.
Agricultural Robot Motion Modes
A silent laboratory extract showing in-place turning and opposite-phase motion on the organic-farming research platform.
3D Gesture Landmark Visualisation
A rotating 3D view of MediaPipe-derived arm and hand landmarks, used to inspect the geometry of a stop gesture before training.
Synchronised Multi-Sensor Capture
Hardware-triggered cameras and LiDAR replayed together in RViz — the panorama and the point cloud advance on the same clock.
05 / PUBLICATIONS
Research, properly connected.
Two peer-reviewed publications, each linked to its publisher, DOI, and indexing records.
GESTRO: Skeleton-Based Dynamic Gesture Recognition for Robot Teleoperation
Journal of Intelligent & Robotic Systems · Volume 112 · Article 51 · Published 16 April 2026
A skeleton-based dynamic gesture recognition system for mobile robot teleoperation, built around an 11-gesture, 3,300-sample dataset and an attention-enhanced LSTM evaluated on a mobile robotics platform.
Contribution: material preparation and data collection; method and analysis.
Human-Robot Interaction in Indoor Mobile Robotics: Current State, Interaction Modalities, Applications, and Future Challenges
Sensors · Volume 26, Issue 6 · Article 1840 · Published 14 March 2026
A comprehensive survey of interaction modalities, applications, user-experience factors, implementation challenges, and future directions for indoor mobile robots in human-centered environments.
First author · Review article · PMID 41902009
06 / EXPERIENCE & QUALIFICATIONS
Research experience, grounded in engineering.
Nine roles across four institutions, from optical networks in India to laboratory robotics in northern Germany.
AUG 2025 — NOWCurrent
Associate Researcher
Human–Robot Interaction for Laboratory Automation
CELISCA — Center for Life Science Automation, University of RostockBuilding the interaction layer for MOLAR, the Mobile Laboratory Robot — an ASTI ProBot L platform serving laboratory stations across a multi-storey building — as part of the ADAM (Autonomous Development of Advanced Materials) project. Verbal and non-verbal interaction in one interface.JUN — JUL 2025
Associate Researcher
Autonomous Cargo Cycle Remote Control
Otto von Guericke University Magdeburg — AuRaHBW with AuRaSysDeveloped a remote control system with an explicit state machine for a human-following autonomous cargo cycle, implemented in ROS 2 and C++.JUN 2024 — MAY 2025
Associate Researcher
Autonomous Mobile Robot for Organic Farming
Otto von Guericke University Magdeburg with aimess GmbHLocal planner, control design, kinematics, and EKF-based sensor fusion for an agricultural robot automating soil-sample collection, validated in Gazebo and in field trials.NOV 2023 — APR 2024
Associate Researcher
Automotive Vision & Sensor Acquisition
Technische Hochschule IngolstadtLed PSF estimation and deblurring for automotive camera health monitoring, and designed the hardware triggering and time-synchronisation system for the LiDAR, camera, GPS, and radar sensor vehicle.MAR — OCT 2023
Master thesis student
Dynamic Hand Gesture Recognition for HRI
Otto von Guericke University MagdeburgDesigned and implemented the gesture recognition and face verification system for the TIAGo robot, tested in Gazebo and RViz and then on the physical platform.SEP 2022 — SEP 2023
Working student
Gesture Recognition Application Development
Otto von Guericke University MagdeburgBuilt the real-time gesture recognition application and GUI that the ROS-based TIAGo robot acts on, including dataset creation and feature engineering.SEP 2021 — JAN 2022
Research assistant
Path Planning & Nonlinear MPC
Otto von Guericke University MagdeburgACADO-based path planning and obstacle avoidance in MATLAB, with a nonlinear model-predictive controller running on the Hamster UGV under OptiTrack motion capture.FEB — MAY 2018
Intern
Metro Data Centre Interconnectivity
NCR Corporation, Mumbai, IndiaField work on a metropolitan data-centre interconnect deployment, covering DWDM, FTTH, GPON, OTN, and SDH transport technologies.JUN — JUL 2017
Intern
Optical Transport Networks
Alcatel-Lucent Enterprise, Gurgaon, IndiaHands-on configuration, testing, and optimisation of DWDM-based optical transport equipment.
M.Sc. Electrical Engineering and Information Technology
Otto von Guericke University Magdeburg · 2019–2023, degree awarded 2024 · final grade 1.8 (German scale, 1.0 = best)
Thesis: Deep Learning-Based Approach for Gesture-Based Human-Robot InteractionB.Tech. Electronics and Communication Engineering
G D Goenka University · 2014–2018 · CGPA 9.34/10, equivalent to 1.3 on the German scale (modified Bavarian formula)
Final-year project: technical report on optical transport networks and SDNAcademic Foundation
State estimation, computer vision, machine learning, nonlinear and optimal control, autonomous mobile robots, FPGA and process control
Selected coursework from the official M.Sc. transcript, plus a minor in finance and international managementRecognition
- 2023Excellence scholarship during the master's thesisOtto von Guericke University Magdeburg
- 2019Gopi Ram Goenka Award at convocationG D Goenka University
- 2014–18Dean's list and merit scholarship, four consecutive yearsG D Goenka University
- 2019Qualified GATE and JEE MainNational engineering entrance examinations, India
TECHNICAL TOOLKIT
Methods and platforms used across the research record.
Robotics
- ROS / ROS 2
- Nav2
- MoveIt
- Gazebo
- RViz
- TIAGo
- OptiTrack
AI & Vision
- PyTorch
- TensorFlow / Keras
- OpenCV
- MediaPipe
- LSTM + attention
- GANs
Control & Estimation
- NMPC
- ACADO
- Kalman / EKF
- MATLAB / Simulink
- Visual SLAM
Engineering
- Python
- C++
- Docker
- Git
- Linux
- Blender
- CARLA
Sensors & Hardware
- LiDAR
- GPS / IMU
- Radar
- Arduino
- Raspberry Pi
- FPGA / PLC
Interfaces
- CAN
- Ethernet
- Serial
- Bluetooth
- TCP/IP
- Wi-Fi
07 / SELECTED CREDENTIALS
Focused technical learning.
A curated subset relevant to robotics, AI, control, and secure engineering. Personal identifiers and certificate scans remain private.
Developing Secure Software
LinkedIn Learning
Verified from supplied certificateROS for Beginners: Basics, Motion, and OpenCV
Udemy
Verify credentialROS 2 for Beginners
Udemy
Verify credentialROS for Beginners
Udemy
Verify credentialIntroduction to Self-Driving Cars
University of Toronto / Coursera
Verify credentialMachine Learning
Sololearn
Verified from supplied certificateData Science
Sololearn
Verified from supplied certificatePython Core
Sololearn
Verified from supplied certificateC++
Sololearn
Verified from supplied certificateC Programming Language
Sololearn
Verified from supplied certificateControl Design Onramp with Simulink
MathWorks
Verified from supplied certificateMATLAB Onramp
MathWorks
Verified from supplied certificatePrivacy note: government IDs, insurance records, transcripts, certificate IDs, signatures, and unredacted documents are intentionally not hosted here.
08 / CONTACT
Let’s make the next idea tangible.
For research exchange, robotics and AI collaboration, technical discussion, or future opportunities.
Follow my research, publications, and engineering work across these verified profiles, or read the full CV.