Robotics

INRIA is one of the research institutes worldwide that specialize in robotics. The results of the research teams that explore this field bring about significant advances in medicine and surgery, concerning automatic systems, command and control processes. In surgery, the trajectory of a robot can be planned then controlled by images acquired before and during the operation, as was demonstrated by the research results of the CHIR initiative.

Aid in surgery is not the only aspect of robotics in medicine. It is now possible to acquire images of the entirety of the digestive tract by ingesting a miniature camera. Soon, it will probably be possible to suppress the apparent movements of an organ, for the benefit of surgeons, movement of the brain or respiration movements in digestive surgery, even heart movements in cardiac surgery.

Robotics research also has promising applications for paralytic patients (development of prostheses, project DEMAR), hearing impaired persons (development of a "talking head", project PAROLE) and renal insufficiency patients (development of dialysis safety systems, DIALHEMO national research initiative).
Project MAIA puts artificial intelligence at the service of remote surveillance of patients that live far away from hospital centers.

DEMAR : Walking and Artificial MovementPractical information

How to restore the mobility of a person whose spinal cord has been damaged when muscles and neuromuscular command remain intact?
Can robotic command of movement provide an answer?

Muscle electro-stimulation through an artificial coordination system - Anthropomorphic biped robot - In situ implant X-ray

This is a crucial question for doctors who treat paraplegic or quadriplegic patients. Project Demar is applying INRIA's expertise in robotics and automatic control to the human biomechanical system in order to understand and model the functioning of muscles and the neuromotor systems that control them.

This work reinforces the approach initiated by Professor Rabischong of the Faculty of Medicine of the University of Montpellier I, who tried to get muscles moving via functional electrical stimulation (FES).

Today, work on biped robotics is bringing about interesting research directions: stimulator implants, closed loop control of muscles under FES, movement analysis.

The project brings multiple skills together. In addition to the Faculty of Medicine of the University of Montpellier I and several INRIA teams, the project is carried out in partnership with CNRS, the University of Montpellier II, and a Danish laboratory (SMI, University of Aalborg) specializing in sensorimotor interactions. A company called MXM is associated with the project for the development of implants.

The primary partners of Demar are the patients themselves since the first criterion for success of the system will be how well it can be tolerated, beyond its quality and reliability. This is why the team has been listening to the needs and wishes of the patients and adapting the project accordingly.

The project in brief

Back to top

PAROLE : Speech Analysis, Perception and Recognition

Can human speech modeling help hearing-impaired persons have easier access to language understanding and practice?

© INRIA Automatic speech recognition via a virtual talking head

There are many situations in which neither sign language, nor lip reading are accessible to hearing-impaired persons, like telephone conversations or a classroom as soon as the teacher turns around to write on the board while speaking. To tackle this problem, the DATHA nonprofit organization (Development of Technological Aids for Auditive Handicapped Persons) intends to develop a “talking head” for complemented spoken language. The talking head is a three-dimensional image of a face that would reproduce lip movements based on the recorded speech signal, complemented by a coding hand. Persons could thus read the lips with the help of information supplied by the hand (vowels given by the position of the hand, consonants by that of the fingers), irrespective of the location and position of the speaker. This ambitious project entails the development of a complex driving system associating speech recognition (to identify the sounds) with a model that is able of linking each sound with lip and face shapes, and its computer graphics “translation”.

The Parole team specializes in the study of spoken communication. It got involved in the project by applying its work in the fields of speech analysis, speech recognition and language models. The scientific stakes are high. It is for example essential to ensure the reliability of the speech recognition system for any given speaker, accent or pronunciation. Similarly, the system must function in spite of various ambient noises. Parole is also participating in the development of the 3D image talking head. To capture the lip movements associated with sounds, the team developed a tracking system using medical images, in partnership with INRIA's ISA team. Parole is currently working on the lip animation model. EDF participates in this project, which is in the framework of the RIAM program (Network for Research and Innovation in Audiovisual and Multimedia) of the Ministry of Research.

The talking head program should also contribute to improving the teaching of French to deaf persons. It is just one example of application of Parole research, among which also are the construction of vocal interfaces for embedded systems, vocal rehabilitation, the improvement of auditive prostheses, language learning, and so on.

The project in brief

Back to top

MAIA : Intelligent Autonomous Machine

Could reliable warning systems adapted to medical surveillance make it possible to remotely monitor patients living far from hospital centers?

The MAIA research team specializes in artificial intelligence. The team is increasingly applying its work to telemedicine projects. For example, MAIA is participating in the development of a warning system associated with peritoneal dialysis, in the framework of the DEPIC project (Early Detection of Cutaneous Infection). Patients suffering from renal insufficiency that follow this type of treatment are equipped with a catheter that enables them to perform peritoneal dialyses at home. The DEPIC sensor will consist of a temperature sensor developed by the Matter Physics Department of INSA Lyons, to be placed around the opening of the catheter, and of an expert system developed by MAIA. The detector will make it possible to forecast the occurrence of a cutaneous infection at this location, based on the temperature readings. This project is in the framework of the RNTS, like the autonomy telediagnosis project to keep elderly persons at home. In the latter case, the goal is to prevent falls by monitoring the balance of elderly people. Various types of sensors are currently under study in partnership with the Technology University of Troyes, the teaching hospital of Nancy and the Nancy Regional Rehabilitation Center. MAIA will then step in to develop a walk model obtained by self-reference, as well as an expert system capable of triggering an alert based on the collected data.

These applications represent an important challenge for MAIA research in artificial intelligence. Indeed, the signals and information recorded in the medical field are quite variable from one individual to another, and there is no reference database. The developed system's own learning capabilities are essential to obtain reliable warnings. MAIA also applies its research to the development of expert systems for anesthesia monitoring, in partnership with Professor Meistelman's team at the teaching hospital of Nancy, who have already experimented a first prototype software, or for transplanted kidney follow-up. The project participates in the developments of the Dialhémo ARC.


The project in brief

Back to top

DIALHEMO : National Research Initiative

Could expert systems enable renal insufficiency patient treated by hemodialysis to follow their treatment at home or in self-dialysis centers with the same level of safety as in medical centers?

In France, 30,000 persons are currently suffering from renal insufficiency and 6,000 new cases are identified each year. These patients must undergo dialysis sessions for which, most of the time, they have to move into a specialized center. Dialysis is a heavy procedure that requires monitoring. For instance, a patient may loose 10 kg in weight during a single session. The total cost of the treatment of this pathology is estimated to amount to 2% of all national health expenses.

The objective of Dialhemo is to develop a remote surveillance and telediagnosis system adapted to renal insufficiency patients treated by hemodialysis, either at home, or in self-dialysis centers. The project is a continuation of the work carried out since 1995 with LORIA (formerly CRIN, Computer Science Research Center of Nancy) that resulted in the Diatélic system that monitors and controls home patients treated by continuous ambulatory peritoneal dialysis.

The current work of Dialhemo is aimed at establishing a data model that will make it possible to optimize the monitoring protocols, in particular by adapting them to each special case of patient-doctor pair. This model will be the basis of an expert system that will monitor the state of health of the patient by estimating his or her dry weight.

The team works with two industry partners, the DIATELIC SA company, a startup stemming from LORIA, and GAMBRO, the world leader on the market of dialysis machines.
Part of the future system developed by Dialhemo will be directly connected to Gambro dialysis machines or to other machines via a generic data exchange protocol.

The surveillance system itself will finally be clinically validated by patients of the ALTIR Saint Jacques Center. A first experimental system is being currently validated with 200 patients.


National Research Initiative in brief