Blog

Can a robot understand its environment? Can a robot be aware of its capabilities and predict the effects of its actions? Can a robot decide the best action to perform in a given situation, or plan the set of actions to achieve a goal? Can a robot adapt the execution steps of an action to the actual scenario? Can the robot have a deliberative and reactive behavior? Can the robot detect the action failures and plan recovery strategies?

UPC is working to answer positively to these questions and make the robot reason on manipulation actions and use this reasoning to adapt its manipulation capabilities to different scenarios. This will allow asking higher level questions related to specific applications, as: Can a robot set the table for breakfast? Can the same robot manipulate fresh food or cables?

In the following article we proposed a smart and adaptive manipulation framework to deal with these issues and make robots more autonomous. The proposal is a knowledge representation and reasoning approach based on ontologies that give the robot situation awareness, domain awareness and execution awareness that allow it to be adaptive at planning, execution and monitoring levels.

👉 Article link: https://upcommons.upc.edu/handle/2117/396236

Coming soon, the first step of this framework will be detailed: how to use knowledge to automatic configure the behavior tree required to execute a given manipulation task.

There are many challenges to be faced and open questions to be answered, and this is the aim of the IntelliMan project.

Hashtags

#roboticmanipulation #reasoning #AI #research #technology #autonomousrobots #awareness #ontologies #assistiverobotics #mobilemanipulators