logo inria

Assistance to the person

Since 2008 we started an effort of long term on assistance to the persons with reduced mobility (elderly and handicapped people). During two years we interviewed field people (subjects, elderly and handicapped association family, helpers, nurses, personnel of retirement home, doctors, local authorities) to identify needs and priorities as well as guidelines which had to guide our developments. It allowed us to identify priorities (what must be made, for whom, when, why):
  • mobility assistance for elderly, caregivers and family. At home autonomous motion or assisting for the motion of a person (i.e. realizing a {\bf transfer} operation) is physically demanding and requires a powerful system. Such an assistance is especially required for getting alone to the toilets both by end-users and the helpers and it must be noted that a contrario most existing mobility assistance devices are very difficult to use in such a narrow room. Demands for outdoor displacement are more confuse but authoritarian navigation system will not be well accepted. Another type of mobility assistance may be required for moving objects (typically helping an elderly to eat alone). Curiously the speed of the existing robots that fulfill this task, that may be seem quite low by outsiders, is judged satisfactory by the end-users but their prices are definitely not
  • ensure safety: this is a high priority demand for all interviewed personnel. The purpose is evidently to ensure that assistance devices may not arm the end-user but also that they must help in an emergency situation. Among all emergency situations the medical community has mentioned that fall was a major issue
  • medical monitoring: the doctors complain about
    • the low frequency with which they get information on the state of health of their patients while their trajectories of life evolve fast. A daily monitoring very close to daily life would allow them for example to discern more easily emergent pathologies
    • the relevance and density of the information they get. While it is necessary that information remains synthetic indicators that are objective, robust and better reflect trends are needed. A special emphasis has to put on the strict privacy that has to ensured for this type of information
  • we need to help also the helpers (family, caregivers) who are sometimes subjected to intense stress (both physical and psychological), what doesn't reinforce their availability for social relation
Besides these priorities we have identified guidelines which have to guide us in our developments:
  • propose system with very low intrusivity: ideally our systems must be invisible and be deployed only on request or in case of an emergency
  • offer systems which are adapted to the user, his trajectory of life and to his environment. Except in particular cases it isn't legitimate to ask for important modifications of the subject environment
  • attend but not take the place: it is always necessary to search the best compromise between the provided help and the necessity to maintain the level of activity at its maximum
  • attend but not replace: the purpose of assistance systems must be to clear time for social relations and to make them easier, not to replace human interaction
  • offer low cost systems not only in the purchase but also for installation and maintenance. The familial environment is often ready to finance at all costs even imperfect devices of assistance which, furthermore don't include recent technological innovations. This leads to expensive devices, the industrial equivalents of which are much less expensive. The cost must also take into account installation and maintenance, that should not be provided by experts but by local workers which are more available in case of problems. Indeed unavailability of an assistance system may have dramatic consequences as well from physical point of view as psychological
  • offer control interfaces that are appropriate to the physical/cognitive aptitude of the users and their variations in time. Furthermore these interfaces, which are used every day and in a repetitive way must allow for the medical monitoring asked by the medical community. We have also to consider hown objects of daily life (and therefore are accepted) may be used for health monitoring, provided a strict respect of privacy
  • offer systems with low power consumption: the aimed public is not the most capable to pay attention to the state of the systems, nor to think of loading / replacing batteries while consequences of a breakdown of energy can be dramatic. Ideally systems should be almost energy autonomous and should not require any voluntary action of the user regarding energy.
  • offer smart systems: in our opinion we are very far from being able to offer an universal system of assistance that will be able to answer all needs (and who would be undoubtedly of a prohibitory cost). For exemple the laws of mechanics plays an adverse role for the transfer operation: a system that has not been specifically designed for this task will be cumbersome and will not satisfy the low-cost, low-intrusivity and low-energy guidelines. Hence before an universal tool may be designed we will have to manage group of assistance agents, that are able to efficiently perform a limited number of tasks (and possibly some other tasks but with a reduced efficiency). These agents will therefore be led to communicate with other agents to exchange information and to collaborate with them (for example in emergency situations). It will however be necessary to pay attention to the confidentiality of data which they exchange, some of them likely medical.
  • think globally for the programming of the assistance systems. We will be dealing with life-critical, distributed, heterogeneous systems evolving in a very variable environment since it must follow the trajectory of life of the user. Hence manual programming of such system isn't possible and we will have to rely heavily on computer science development for such programming
  • give a sense of responsibility to the users and familial surrounding: an assistance device will never be able to be 100% reliable. Applying without judiciousness a principle of precaution leads to concentrate industrial efforts towards solutions without risk (some of which furthermore allows for large financial gains) but which aren't the most appropriate for the end-users.
Another issue is that in this domain experimental validation is absolutely compulsory. This raises two major problems:
  • legal issues : in France experiments involving humans are strictly ruled: a formal procedure involving the Comit\'e de protection des personnes (CPP), roughly the equivalent of an ethical research committee, has to be respected and is in our experience demanding,
  • experimental test in realistic situation are required and consequently a specific test infrastructure has to be created.
To answer these priorities and to respect these guidelines we have started developments according to specific axis:
  • to develop very strong collaboration with the medical community (for example with Nice hospitals)
  • to have our own test environment which allows us to test our assistance systems in realistic conditions, environment which should be shared with others resources centers. For these purposes we have implemented a complete flat with bedroom, automated kitchen, corner meal and rehabilitation zone
  • reduce costs by using only standard hardware, reliable and quickly available.
  • offer design methodologies which take as inputs the user abilities, his environment and a possibly limited list of components and that provide as output modular and adaptable solutions with guaranteed performances, in spite of unavoidable uncertainties. Furthermore these methodologies also participate in the reduction of costs by reducing the level of expertise that is necessary for installation.
  • include since the very beginning of the design that any assistance system can play a role in the medical monitoring
  • include at best the most innovative solutions in computer science which allow to interface very easily with sensors while having a very low energy consumption (or even almost none with energy harvesting systems. For example we are using extensively Phidgets, fit-pc or Arduino.
Of course we don't pretend to be able to reach all these targets alone and in a few years. This is why we have initiated the proposal of the Large Scale Action PAL, officially created in 2011, and that we try permanently to tie collaboration, including one that are far away from the robotics domain. For example we have collaboration with the project-teams INDES (programming), STARS (monitoring of activities by vision), REVES and V4R (virtual reality). We also think that human and social sciences should play an essential role in this activity, for example for social acceptance, for interfaces and on the strengthening of social relations but also on ethical problems, which cannot be dismissed for such a research topic. We however have began exploring possibles and developing prototypes which are summarily introduced in this page .