Abstract
The improvement of odometry systems in collaborative robotics remains an important challenge for several applications. Social odometry is a social technique which confers the robots the possibility to learn from the others. This paper analyzes social odometry and proposes and follows a methodology to improve its behavior based on cooperative reputation systems. We also provide a reference implementation that allows us to compare the performance of the proposed solution in highly dynamic environments with the performance of standard social odometry techniques. Simulation results quantitatively show the benefits of this collaborative approach that allows us to achieve better performances than social odometry.
Highlights
This paper proposes a new approach for collaborating purposes in a swarm of robots working together to achieve a goal
Trust information could be disseminated faster and the whole system performance might be improved as well. Based on these previous ideas our trust algorithm will be defined as follows: The inputs for our algorithm will be: (i) the distance traveled since the last known location, so we can keep the advantages of the classical social odometry approach; (ii) the category or type of robots in the system, so we can introduce an a priori knowledge but in a simpler way than using other common techniques; and (iii) the ratio distance divided by number of round-trips, so we will have an estimate of the individual performance
In the horizontal axis, we will display a boxplot for each of the studied odometry techniques
Summary
This paper proposes a new approach for collaborating purposes in a swarm of robots working together to achieve a goal. The importance of social odometry lies on the fact that the swarm (the collectivity) allows the robots to collaborate to achieve a common objective because the individuals are working together. Odometry is probably the most used as it provides easy and cheap real time position information by the integration of incremental motion information over time This integration causes an accumulation of errors during the movement of the robot, and this can be a great drawback in some robotic applications, such as foraging, where the robots have to find, select and exploit resources from unknown locations. We work with a classical swarm foraging scenario: a number of resource items (usually called “prey”) are randomly scattered in the arena In this context, robots search and retrieve those resource-items back to a specific place (usually called “nest”).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
