A POSTDOCTORAL POSITION in Robust control architecture for Mobile Manipulator Robots (MMR)

About this position

The ImViA laboratory (Imaging and Artificial Vision, UR 7535) is a research unit of the University of Burgundy Europe (UBE), recognized for its work at the interface between computer vision, image processing and control of physical systems. It brings together several thematic teams, including a team dedicated to vision-assisted control of physical systems, developing innovative approaches where visual perception is integrated as an active sensor in the control loop of complex robotic systems.

This team conducts research on robust control, state estimation by vision, visual servoing and machine learning applied to cyber-physical systems. It has instrumented experimental robotic platforms (cameras, mobile robots, manipulator arms) enabling real-condition validation of the developed algorithms.

The MMR project (Robust control Architecture for Mobile Manipulator robots) aims to propose solutions that can address a major challenge of modern robotics: enabling mobile manipulator robots to operate safely, robustly and efficiently in dynamic and weakly structured environments. These systems, combining an autonomous mobile platform and an articulated manipulator arm, find strategic applications in infrastructure maintenance, intelligent logistics, inspection, rescue in disaster areas and precision agriculture.

The originality of the MMR project lies in the synergistic integration of robust predictive control (Robust/Stochastic MPC), observers coupled with Physics-Informed Neural Networks (PINN) and Control Barrier Functions (CBF), with active exploitation of vision as a sensory modality for environment perception and real-time state estimation.

Research domain:

Control of physical systems, mobile robotics and manipulation, robust control and real-time constrained optimization, computer vision applied to robotics.

Mobile manipulator robots represent a major evolution of modern robotics. Intrinsically nonlinear, strongly coupled and underactuated, they remain sensitive to external disturbances (ground irregularities, load variations) and internal uncertainties (variable parameters, approximate models), making their control particularly challenging under real-time constraints.

Three locks condition their effective deployment:

Dynamic lock

: the nonlinear coupling between the mobile base and the manipulator arm complicates control design and stabilization, particularly during simultaneous movements.

Robustness lock

: disturbances and uncertainties degrade precision and stability. Existing approaches do not allow robustness and energy efficiency to be guaranteed simultaneously [2].

Safety lock

: formally guaranteeing collision avoidance while satisfying multi-objective constraints in real time. Predictive controllers (MPC) struggle to integrate formal safety and robustness to unbounded uncertainties; pure machine learning lacks formal stability guarantees for critical applications [3].

The MMR project proposes a control architecture based on three principles:

Robust predictive control (Robust/Stochastic MPC)

to anticipate dynamic interactions and manage uncertainties proactively [4];

Real-time estimation

of disturbances and uncertainties by robust observers coupled with Physics-Informed Neural Networks (PINN), enriched by vision data (obstacle detection, pose estimation, scene reconstruction);

Formal safety guarantee

via Control Barrier Functions (CBF) integrated into optimization constraints [5].

References: [1] Khan & Thomas, 2026 — [2] Gao et al., 2026 — [3] Heins & Schoellig, 2023; Ames et al., 2019 — [4] Guo et al., 2026 — [5] Thirugnanam et al., 2022.

Job description and activities

Develop and validate a robust control architecture for mobile manipulator robots integrating MPC, adaptive observers and CBF, with exploitation of vision as a sensor in the control loop.

Design and implement robust observers coupled with Physics-Informed Neural Networks (PINN) for real-time estimation of disturbances, uncertainties and system state, leveraging available visual data.

Integrate Control Barrier Functions (CBF) as optimization constraints to formally guarantee collision avoidance, exploiting visual perception of the environment.

Carry out simulations under MATLAB/Simulink, then experimentally validate the developed algorithms on the instrumented robotic platforms of the ImViA laboratory.

Actively contribute to the scientific dissemination of results: writing articles in rank A international journals and presenting at national and international conferences.

Participate in project meetings, occasionally supervise master's students and contribute to the scientific life of the ImViA laboratory.

Relation scope

:

The postdoctoral researcher will work in close collaboration with:

The MMR project leader within the ImViA laboratory (direct scientific supervision)

The members of the Physical Systems Control and Vision team at ImViA (researchers, PhD students, engineers)

The robotic and vision experimental platforms of the laboratory