DYROLL · Behavioral planning for dynamically dexterous robots using constrained intuitionistic linear logic
FP6 — Marie Curie Actions (Human Resources and Mobility)
- Duration
- 2007-10-01 → 2009-09-30
- EU contribution
- €80,000
- Participants
- 1
- Scheme
- IRG
Lines connect the coordinator with its partners. CORDIS does not always give exact coordinates for projects before 2014. These points are placed at city or country level.
Results in brief
Final Activity Report Summary - DYROLL (Behavioral Planning for Dynamically Dexterous Robots using Constrained Intuitionistic Linear Logic)
The primary objective of this project was to investigate the use of linear logic for robotic planning problems. Three key contributions were made in the scope of the project. First, continuous constraint expressions were incorporated into the existing language of linear logic to make it sufficiently expressive for encoding properties of dynamically dexterous robot platforms. Second, a sound and complete proof theory was developed for the full fragment of linear logic, incorporating focusing and resource management for overall efficiency and tractability. This directly led to the implementation of an associated theorem prover, whose annotated proof constructions can be used to obtain provably correct behavioural plans for robots. Finally, an improved version of the hexapedal RHex robot platform was developed to support subsequent experiments in applying the ideas developed in this project to real-world robotic planning problems such as those encountered in search and rescue applications. These three components tightly connect together towards a robust planner for autonomous robot behaviour in the presence of fast, dynamic behaviours that cannot be adequately described by discrete symbols alone.
Data: CORDIS, © European Union
Project objective
We propose to augment intuitionistic linear with two new connectives to incorporate constraints drawn from a possibly continuous domain into linear logic expressions. The resulting formal language is called Constrained Intuitionistic Linear Logic (CIL L). Our ultimate goal is to capture physical properties of dynamically dexterous robotic systems within such a constraint domain while reasoning with changing state information through the representational strength of linear logic.The resulting integration of linear, discrete reasoning with an encapsulation of dynamic properties of a system will be used to achieve behavioural planning of a fast, autonomous legged robot in complex outdoor environments. To this end, we will first formulate physically relevant semantics for CILL and prove the completeness and soundness of an associated proof theory. We will then adapt a number of existing solvers for real number and set constraints to adequately capture expressions that may arise from behavioural sequencing for dynamical systems. Finally, we will use the resulting language to encode behavioural sequencing problems for a legged robot and experimentally verify their relevance and correctness.
Original text from CORDIS.
Participants
- BILKENT UNIVERSITESI · ANKARACoordinatorCity levelTürkiye
Links
Data: CORDIS, © European Union