Assisted Navigation for Persons with Reduced Mobility: Path Recognition Through Particle Filtering (Condensation Algorithm)

作者：Grasse Regis; Morere Yann^*; Pruski Alain

来源：Journal of Intelligent & Robotic Systems, 2010, 60(1): 19-57.

DOI：10.1007/s10846-010-9406-y

• Assistance to navigation
• Particle filtering
• Condensation algorithm
• Course modelling
• Recognition
• Smart wheelchair

In the few past decades, several international researchers have worked to develop intelligent wheelchairs for the people with reduced mobility. For many of these projects, the structured set of commands is based on a sensor-based command. Many types of commands are available but the final decision is to be made by the user. A former work established a behaviour-based multi-agent form of control ensuring that the user selects the best option for him/her in relation to his/her preferences or requirements. This type of command aims at "merging" this user and his/her machine-a kind of symbiotic relationship making the machine more amenable and the command more effective. In this contribution, the approach is based on a curve matching procedure to provide comprehensive assistance to the user. This new agent, using a modelization of the paths that are most frequently used, assists the user during navigation by proposing the direction to be taken when the path has been recognized. This approach will spare the user the effort of determining a new direction-which might be a major benefit in the case of severe disabilities. The approach considered uses particle filtering to implement the recognition of the most frequent paths according to a topological map of the environment.

• 出版日期2010-10