«Back to research topics list

Applications in Robotics

Inverse Kinematics Problem
Inverse Kinematics (IK) problem concerns finding such a configuration of an articulated robotic arm that satisfies certain constraints regarding its position and orientation. My interests focus on solving this problem in a constrained, dynamic case, i.e. guiding the arm in an arena containing moving obstacles. 		Benchmark I


Benchmark II


Benchmark III



Examples of arenas used for testing the algorithms in the paper A Predictive Evolutionary Algorithm for Dynamic Constrained Inverse Kinematics Problems (click to open a larger version)



Infeasibility Driven Evolutionary Algorithm with the Anticipation Mechanism for the Reaching Goal in Dynamic Constrained Inverse Kinematics
The research presented in a paper at the GECCO 2015 Conference.

Highlights of the paper:
  • A new algorithm mIDEA-ARIMA is used which composes the population from three fractions: exploring, exploiting and anticipating.
  • The mIDEA-ARIMA algorithm improves on the IDEA-ARIMA algorithm presented in a paper at the IDEAL 2011 Conference.
  • Predictions of the most probable future landscapes are used for acting prior to incoming changes.
  • An additional optimization criterion for minimizing the displacement between joint angles con gurations obtained in the consecutive time steps is introduced, which assures a smooth motion of a robotic arm.



Examples of arenas used for testing the new algorithm.



Multiobjective Dynamic Constrained Evolutionary Algorithm for Control of a Multi-Segment Articulated Manipulator
The research presented in a paper at the IDEAL 2014 Conference.

Highlights of the paper:
  • New algorithm that handles many-segment manipulators (up to 100 segments).
  • Improved genetic operators dedicated for the Inverse Kinematics problem.
Video of the test for 10-segment manipulator (click to view).
Video of the test for 100-segment manipulator (click to view).



A Predictive Evolutionary Algorithm for Dynamic Constrained Inverse Kinematics Problems
The research presented in a paper at the HAIS 2012 Conference.

Highlights of the paper:
  • Proposing a new evolutionary algorithm based on the Infeasiblity Driven Evolutionary Algorithm that uses ARIMA modelling for prediction of future values of objective functions and constraints in a dynamic environment.
  • Defining an objective function that includes convex hulls of the obstacles in the calculation of distance to the target (see the figure on the right).
  • Defining constraint violation measures that increase the selective pressure towards outside of the obstacles.
  • Comparing the performance of the proposed IDEA-ARIMA algorithm with the standard IDEA algorithm (without a prediction mechanism).
The calculation of the distance to the target taking into account the path around the obstacles (click to open a larger version).

δ1L, δ2R and δ3L are distances around the obstacles.

d1, d2, d3 and d4 are distances along the straight line to the target.
«Back to research topics list