Projects

EnhancINg Surgery with deeP lEarning-controlled Continuum roboTs (INSPECT)

Developing reliable and explainable Artificial Intelligence (AI) approaches is of paramount importance especially in clinical applications. The wide adoption of robots for surgery is established, with a market of $16 billion in 2024. This domain was marked by the introduction of continuum robots in the last decades, a break-through in the robotics paradigm, allowing for inherently safe, miniaturized, and deformable systems, able to access and navigate through complex anatomy and closely reach confined therapeutical targets. Their wide spread is yet limited by the ability to localize and control such small devices in medical images. INSPECT will provide AI (deep/reinforcement learning) methods, combined with recent sophisticated robot mathematical models, in order to precisely estimate the entire shape of continuum robots and assist surgeons to accurate control their motion. INSPECT will aim for advances in AI-based surgical robots: less invasive, safe, reliable, and accurate.

Patient-Specific Continuum Robots for Assistance to Percutaneous Nephrolithotomy (CORNEL)

Lithiasic disease is often treated through an invasive intervention called percutaneous nephrolithotomy (PCNL), allowing to gain access directly to the kidney to extract the stones, which requires the use of straight, stiff instruments. Such procedure implies limited dexterity and access, increasing risks of bleeding and residual stones, intensified by the complex anatomy of the organ. Major scientific challenges should be addressed in order to provide an efficient and safe alternative. Indeed, continuum robots - slender systems, inherently miniaturized and deformable - are very relevant candidates for intervention assistance, as to adapt to inter-patient variability, the need to explore the kidney, and proximity to fragile tissue and vital organs. Based on an accurate modeling approach, CORNEL project will develop design, perception, and control methods for continuum robots. This project considers a more general framework in fundamental research of deformable robots, evolving in a deformable environment, and interacting with deformable tissue and organs that are relatively mobile, therefore constituting a disruption of the established paradigm of standard rigid robots.

Robots for Real World Interaction (BOOT)

BOOT project aims to develop interdisciplinarity and scientific innovation in the field of robotics interacting with the real world in a strategy that departs from the current vision of robotics and its limitations. Moreover, this project aims to position Grenoble as one of the major players in robotics interacting with the real world, visible at the national and international levels.

Challenges

The current societal and economic challenges, strongly linked to the ethical balance of this new human-robot ecosystem, are to open the real world to robots in a harmonious synergy. It is a very visible and immediate challenge in all fields of robotics: industrial robotics, service robotics, medical robotics and social robotics. It is clear that no robot currently designed can yet satisfactorily meet this challenge, due to the complex, changing and ill-defined nature of the interaction taking place in the applications.

Interdisciplinarity

In this project, we adopt a resolutely multidisciplinary approach to robotics based on Grenoble’s robotics communities in Mathematics, Information, and Communication Sciences (MICS) and Social Sciences and Humanities (SSH) in a strategy that departs from the current vision of robotics and its limitations. With the opportunity of this common framework of cooperation federating Grenoble's strengths, robotics will benefit from the integrative emergence of the competences of each of the actors, each of whom is highly recognized in their specialty.

A transformative project

A cooperative and visible Grenoble community will be built in the field of robotics interacting with the real world by federating Grenoble's robotic skills in the engineering sciences of automation, mechatronics, signal processing, image processing, and computer science, and in the human and social sciences of cognition, social psychology, work psychology, neurobiology, automatic language processing, and ergonomics.

National and international reach

Thanks to recent technological progress and the cultural evolution of our societies, robots are more and more present in our daily life (automotive industry, service robots, medical robotics, assistant robots, etc.) and share their environments and tasks with humans. Robotics is considered as an imminent technological and societal revolution: the "France 2030" investment plan has announced an amount of 800 million Euros for robotics.

For a better visibility of Grenoble

This Cross Disciplinary Program (CDP) aims to position Grenoble as one of the major players in robotics interacting with the real world, visible at the national and international levels. Indeed, there are few places in France, and even internationally, that have such a multidisciplinary community with the skills and the will to address the major societal and economic challenge that is robotics interacting with the real world in which humans evolve.

Neurotechnologies for functional rehabilitation (Grenoble-Neurotech)

Grenoble-Neurotech project is a Cross-Disciplinary Program (CDP) that aims to develop neurotechnologies for functional rehabilitation. The project also explores neurofeedback approaches for psychiatric disorders. A tight collaboration with philosophers accompany these scientific developments to reflect on the ethical implications of neurotechnologies.

What is a Cross-Disciplinary Program (CDP)?

Launched in 2016 as part of the Université Grenoble Alpes (UGA) Initiative of Excellence (IDEX), Cross-disciplinary Programs (CDPs) are ambitious cutting-edge research projects built at the confluence of several disciplines to advance science, innovation and provide answers to the major challenges of the 21st century. These projects, which are resolutely interdisciplinary, give pride of place to human, political and social sciences, with a strong commitment from the scientific community of the SHS (Human and social sciences) and PSS (Social Sciences department) clusters. Here again, CDPs allow the UGA to become a leader in interdisciplinary themes such as the links between beauty and health, research-creation, heritage science and environmental modeling.

Grenoble-Neurotech project

The Grenoble-Neurotech project is one of the 21 CDPs which aims to develop neuroprostheses allowing paralyzed people to compensate - at least partially - for the loss of motor functions such as the control of their upper limbs or speech. It combines new implantable neural interfacing solutions with methods and tools for real-time processing of neural signals. The project also explores how these rehabilitation strategies can be extended to closed-loop therapeutic approaches in certain disabling psychiatric diseases such as obsessive-compulsive disorder. This research is conducted in conjunction with a reflection on the ethical and societal implications of these neurotechnological developments.

Cosserat Rod Theory for Slender Robots (COSSEROOTS)

Robotics is today experiencing a paradigmatic revolution. The “stiffer is better” of our rigid robots is challenged by a new generation of robots with controlled deformations of finite amplitudes. This is the case of trunk-robots in soft robotics, continuum robots in medical robotics or hyper-redundant eel/nake-robots in bio-inspired robotics. In contrast to the rigid elder sibling, this novel robotics lacks a solid and unified corpus of modelling tools for control. Building on a novel parametrization of the Cosserat theory of rods, COSSEROOTS aims at producing for the first time a comprehensive and unified corpus of methods and models devoted to the control of slender robots with actuated large deformations. Beyond modelling, COSSEROOTS will use these novel modelling tools to address challenging control problems in the fields of soft, medical, and bio-inspired robotics.

Continuum Robots in Clinical Applications: Prototypes and Simulation Environment

Medical Interventions (surgery, interventional radiology, radiotherapy) can provide a significant boost for progress in terms of patient-specific optimal planning and performance. To fulfill patient’s demand for Quality, Senior Operators demand to see beyond the immediately visible, to be assisted in their real-time vital decisions and to accede to enhanced dexterity, while junior operators request to “learn to fly” before being left alone, and Public Health Authorities and companies require demonstration of the Medical Benefit of innovations.

Computer Assisted Medical Interventions: A new perspective through the CAMI LABEX

The Computer Assisted Medical Interventions (CAMI) LABEX strategic vision is that an integrated approach of medical interventions will result in a breakthrough in terms of quality of medical interventions, demonstrated in terms of medical benefits and degree of penetration of CAMI technology in routine clinical practice.

The Mission of CAMI LABEX

CAMI LABEX proposes :

• to offer the operator the possibility to see beyond the immediately visible by innovative fusion of multimodal data obtained by novel or classical sensors.

• to offer assistance to real-time decision-making through high-level planning and monitoring of the intervention.

• to offer the operator a new dimension in intervention performance with miniaturized robots and solutions for augmented dexterity.

• to reduce the learning curve by offering User-centered learning strategies exploiting the educational potentialities of CAMI technologies

• to develop and validate an adapted methodology for the demonstration of the Medical Benefit of CAMI techniques.

Continuum Robots in Clinical Applications: Prototypes and Simulation Environment

We are developing a continuum robotic platform by building millimeter-size robot prototypes for demonstration and investigations, as well as a simulation environment. The goal is to gather scientific researchers, clinical partners, and industrial partners together in a unique framework for know-how and know-why exchange. The Continuum Robotic Assisted Medical Interventions ToolKit (CoRAMI-TK) will be part of the ECCAMI (Excellence Center for Computer Assisted Medical Interventions) platform and integrated in CamiTK (Computer Assisted Medical Intervention Tool Kit) framework.

Control of Millimeter-size Continuum Robots for Medical Applications:

My postdoctoral research was essentially focused on feedback control of miniaturized continuum robots (diameters below 10 mm). At this small scale, continuum robots are suitable for a wide range of medical applications where anatomy constraints, narrow paths, and safety requirements prevent standard rigid-linked robots to access and operate. Thus, I am interested in improving functionalities of medical robots by increasing the number of continuum arms accessing from a single endoscope's working channels. This concerns the integration of two collaborative Concentric Tube Continuum Robots and their automatic motion coordination. To this aim, the collaborative continuum arms were kinematically modeled as a single structure interacting with a virtual object. This approach allows to take advantage of the resulting redundancy in the combined system. Therefore, a suitable redundancy resolution allowed to automatically control the relative distance between the collaborative arms' end-effectors while following complex trajectories in 3D space [RAL-2018a].

Another research topic is the description of continuum robots' behavior for optimization purposes and/or for control strategies. We assessed the performances of two of the most widely used kinematic models, namely beam mechanics (BM) and Cosserat rod (CR) theory, both qualitatively and quantitatively. These approaches were considered in modeling a magnetic extensible tendon-actuated robot, with three segments. Our findings include a slight preference for CR in contrast to BM in terms of accuracy: shape error and Euclidean distance error at every segment end. Potential applications include design optimization and kinematic analysis. However, BM showed much higher update rate (up to 1000 times faster than CR model), which is more suitable for real-time applications, such as feedback control [RAL-2019].

I also contributed to improving motion generation abilities of Tendon-Actuated Continuum Robots (TACR). While the standard TACR are actuated by tendons (generally 3) routed parallel to the backbone (robot's center-line), the studied approach consisted of routing an additional tendon helically. The benefits of the additional actuation tendon were quantified in terms of reachable workspace and motion generation abilities. We showed that the workspace can be increased up to 400% and that twisting motion can be provided, added to the regular bending [IROS-2017, video included on IEEE Xplore].

My publications at the Laboratory for Continuum Robotics are listed here: Publications-LKR.