H. Levent Akin

RoboCup Rescue Robot and Simulation Leagues

The RoboCup Rescue Robot and Simulation competitions have been held since 2000. The experience gained during these competitions has increased the maturity level of the field, which allowed deployin ...

Estimating Distributions in Genetic Algorithms

The canonical operators of genetic algorithms, i.e., mutation and crossover, have nondeterministic effects on the population.They use information from only one or two fit individuals and risk deforming the chromosomes of fit individuals and cause an interruption in the progression. Estimation of Distribution Algorithms (EDAs) use probabilistic models rather than mutation and crossover, to guide the progression of genetic algorithms by placing a density over all fit individuals and sampling from this density. EDA therefore makes better use of the fitness information of the previous generation and promise faster convergence without losing any schemata. We consider parametric, nonparametric, and semiparametric models for density estimation in the EDA template with continuous genes. We compare these methods with standard backpropagation and GA proper in the problem of training a multilayer perceptron which is a complex nonlinear estimator. Our results indicate that our algorithms perform definitely better than the proper genetic algorithm (GA) on every problem and can find better solutions than those of backpropagation in training a MLP.

An SSVEP based BCI to control a humanoid robot by using portable EEG device

Brain Computer Interfaces (BCIs) are systems that allow human subjects to interact with the environment by interpreting brain signals into machine commands. This work provides a design for a BCI to control a humanoid robot by using signals obtained from the Emotiv EPOC [11], a portable electroencephalogram (EEG) device with 14 electrodes and sampling rate of 128 Hz. The main objective is to process the neuroelectric responses to an externally driven stimulus and generate control signals for the humanoid robot Nao accordingly. We analyze steady-state visually evoked potential (SSVEP) induced by one of four groups of light emitting diodes (LED) by using two distinct signals obtained from the two channels of the EEG device which reside on top of the occipital lobe. An embedded system is designed for generating pulse width modulated square wave signals in order to flicker each group of LEDs with different frequencies. The subject chooses the direction by looking at one of these groups of LEDs that represent four directions. Fast Fourier Transform and a Gaussian model are used to detect the dominant frequency component by utilizing harmonics and neighbor frequencies. Then, a control signal is sent to the robot in order to draw a fixed sized line in that selected direction by BCI. Experimental results display satisfactory performance where the correct target is detected 75% of the time on the average across all test subjects without any training.

Sound and complete qualitative simulation is impossible

State-of-the-art qualitative simulators (for instance, QSIM) are known to be sound; no trajectory which is the solution of a concrete equation matching the input can be missing from the output. A simulator which is seen to be incomplete, that is, which produces a spurious prediction for a particular input, can usually be augmented with an additional filter which eliminates that particular class of spurious behaviors, and the question of whether a simulator with purely qualitative input which never predicts spurious behaviors can ever be achieved by adding new filters in this way has remained unanswered until now. We prove that such a sound and complete qualitative simulation algorithm does not exist.

A Robotic Fitness Coach for the Elderly

The ultimate goal of ambient assisted living is to help elderly people live a healthy life in the convenience of their homes by making more intelligent technology bring them a set of required assistive tools. In this paper we describe a robotic fitness coach that learns a set of physical exercises from a professional trainer, and assists elderly subjects in performing these gestures. The gestures were selected from an actual training programme at an elderly care home. When demonstrating gestures, the robot performs the learned gestures to the best of its abilities, and while monitoring the elderly subject with an RGB-D camera, provides verbal guidance to complement the visual display, correcting gestures on the fly. We provide a detailed description of the training programme, the gesture acquisition, replication and evaluation algorithms, our solution to the robot stability problem, and a set of preliminary user tests to validate our approach.

Task Refinement for Autonomous Robots using Complementary Corrective Human Feedback

A robot can perform a given task through a policy that maps its sensed state to appropriate actions. We assume that a hand-coded controller can achieve such a mapping only for the basic cases of the task. Refining the controller becomes harder and gets more tedious and error prone as the complexity of the task increases. In this paper, we present a new learning from demonstration approach to improve the robots performance through the use of corrective human feedback as a complement to an existing hand-coded algorithm. The human teacher observes the robot as it performs the task using the hand-coded algorithm and takes over the control to correct the behavior when the robot selects a wrong action to be executed. Corrections are captured as new state-action pairs and the default controller output is replaced by the demonstrated corrections during autonomous execution when the current state of the robot is decided to be similar to a previously corrected state in the correction database. The proposed approach is applied to a complex ball dribbling task performed against stationary defender robots in a robot soccer scenario, where physical Aldebaran Nao humanoid robots are used. The results of our experiments show an improvement in the robots performance when the default hand-coded controller is augmented with corrective human demonstration.

Push-manipulation of complex passive mobile objects using experimentally acquired motion models

In a realistic mobile push-manipulation scenario, it becomes non-trivial and infeasible to build analytical models that will capture the complexity of the interactions between the environment, each of the objects, and the robot as the variety of objects to be manipulated increases. We present an experience-based push-manipulation approach that enables the robot to acquire experimental models regarding how pushable real world objects with complex 3D structures move in response to various pushing actions. These experimentally acquired models can then be used either (1) for trying to track a collision-free guideline path generated for the object by reiterating pushing actions that result in the best locally-matching object trajectories until the goal is reached, or (2) as building blocks for constructing achievable push plans via a Rapidly-exploring Random Trees variant planning algorithm we contribute and executing them by reiterating the corresponding trajectories. We extensively experiment with these two methods in a 3D simulation environment and demonstrate the superiority of the achievable planning and execution concept through safe and successful push-manipulation of a variety of passively mobile pushable objects. Additionally, our preliminary tests in a real world scenario, where the robot is asked to arrange a set of chairs around a table through achievable push-manipulation, also show promising results despite the increased perception and action uncertainty, and verify the validity of our contributed method.

RoboCup 2004 Competitions and Symposium: A Small Kick for Robots, a Giant Score for Science

RoboCup is an international initiative with the main goals of fostering research and education in artificial intelligence and robotics, as well as of promoting science and technology to world citizens. The idea behind RoboCup is to provide a standard problem for which a wide range of technologies can be integrated and examined, as well as being used for project-oriented education, and to organize annual events open to the general public, at which different solutions to the problem are compared. The eighth annual RoboCup -- RoboCup 2004 -- was held in Lisbon, Portugal, from 27 June to 5 July. In this article, a general description of RoboCup 2004 is presented, including summaries concerning teams, participants, distribution into leagues, main research advances, as well as detailed descriptions for each league.

An autonomous robotic exercise tutor for elderly people

Ambient assisted living proposes to utilize technological solutions to sustain the well being of elderly people. In accordance with the vision of successful aging, we describe in this study an autonomous robotic exercise tutor for elderly people. The robot learns a set of physical exercises from a human demonstrator in an imitation framework, and performs these motions in an exercise scenario, while monitoring the elderly person to provide verbal feedback. We developed an exercise program in collaboration with a nursing home, and tested our system in a real world scenario with visitors of a day care center, over multiple sessions. We provide a detailed description of the system implementation, as well as our observations for the exercise program. For the study held in the day care center, video annotations and user self-assessments are evaluated to measure the overall performance of the system and to validate our approach. The analysis revealed that elderly people can successfully exercise with the assistance of the robot, while staying engaged with the system over multiple sessions.

All Bids for One and One Does for All: Market-Driven Multi-agent Collaboration in Robot Soccer Domain

In this paper, a novel market-driven collaborative task allocation algorithm called “Collaboration by competition / cooperation” for the robot soccer domain is proposed and implemented. In robot soccer, two teams of robots compete with each other to win the match. For the benefit of the team, the robots should work collaboratively, whenever possible. The market-driven approach applies the basic properties of free market economy to a team of robots for increasing the profit of the team as much as possible. The experimental results show that the approach is robust and flexible and the developed team is more succcessful than its opponents.