Robots - superheroes of cooperation and surveillance
Robot Cop Guard - Inria
Strange creatures answering to the name of RCG have just invaded the Inria Plateau at EuraTechnologies. These three cooperative and surveillance super robots are the main features in the new demonstration area that has recently been set up in the Inria Plateau exhibition space. They illustrate the research work of the Inria Non-A team (a joint venture with the École Centrale de Lille, the CNRS and the University of Lille 1) towards developing methodologies and software for robotic environments.
The robots move along a predefined path (from point A to point B). Their surveillance mission for the area involves moving about while avoiding obstacles and detecting any anomalies. To complete this task, the robots will execute a precise sequence, divided into two key stages:
- Mapping and localization
Each robot has already mapped out the area where it maneuvers and locates itself according to its environment (obstacles, walls, etc.).
- Planning and monitoring
To follow the path, the robots plan their journey by monitoring and optimizing their movements.
These two major stages call for the development of specific and innovative algorithms applied to the world of robotics.
Simultaneous localization and mapping (SLAM) is nothing new in the field of robotics. SLAM involves creating a map of your environment and being able to pinpoint your location in it. Today it is quite common to program a robot to create a map, but the difficulty lies in knowing what its environment looks like and where the robot is located, all in real time. The latest feat for the Non-A research team is the development of algorithms that make collaborative SLAM possible. Several robots join forces on an initial common surveillance mission ("synchronize watches" mission-impossible style) and locate each other in the area to be mapped. The results of their analysis are then shared to create a very precise map.
Once this collaborative surveillance stage is complete, the second stage will again enable very precise methodologies and algorithms to be implemented, which are relevant to the robots' mission, namely moving from point A to point B.
In this well-identified space, the robots must plan their movements and monitor their environment in real time. The algorithms developed by the researchers will enable tasks to be completed with extreme precision, while taking into account environmental constraints (various obstacles, etc.) and all within a very limited time frame. This precision in a very small area will then be recalculated as the robot moves along. In any given period, the algorithms will recalculate the data to optimize movement by making choices that will best help to complete the mission (for example accelerating or reducing speed to save battery).
Optimization is based on the calculation of a trajectory between an initial configuration and a given configuration on arrival. To complete this stage, the researchers used and developed algebraic tools that are designed to overcome a number of constraints, such as avoiding obstacles or collisions between robots, keep a maximum distance between members of a fleet, or manage energy consumption. Planning the trajectory uses a strategy over a short time frame while taking the robot's immediate environment into account, which means better performance and less time spent on calculations. The addition of a regulator to vehicle movement ensures that the optimal trajectory generated in the previous optimization stage is properly followed.
The robots are thus very precise in their journey and include a control and monitoring mode, which is essential for avoiding obstacles.
In the field of robotics, these libraries of algorithms have a range of potential applications: building surveillance, transporting objects in warehouses, automated transport, autonomous vehicles, drone fleets, collaborative robotics... the list goes on.
Today, the team continues to develop real-time algorithms whose rapidity of detection and response is a major asset.