Abbas AbdolmalekiFormer PhD Student
Abbas Abdolmaleki joined the IEETA and IRIS Labs as a Ph.D. student in September 2011 to join the Robot Soccer Project under the supervision of Luis Paulo Reis and Nuno Lau. During his PhD, Abbas is working on stochastic search methods with robotics application in mind. Abbas also has participated in several robotic projects, and has achieved several awards in international competitions, as for example RoboCup. Currently he is the team leader of FC Portugal soccer humanoid robots simulation project.
Before his Ph.D, Abbas completed his Master Degree in Computer Engineering filed of Artificial Intelligence at the University of Isfahan. His thesis entitled “Trajectory generation for omnidirectional biped robot walking using machine learning methods” was written under the supervision of Nasser Ghasem-Aghaee. Besides, he worked on rescue robots and humanoid soccer robots. Current version of his Curriculum Vitae is available here, and his full list of publications and corresponding bibtex files can be found on his Google scholar account. A few source codes of his projects can be accessed from his Github account. And some of the videos of his projects can be found here.
My research is inspired by the aim of creating intelligent systems, with an emphasis on self-learning, which gives adaptation capabilities in unknown environments. In particular I investigate the challenges behind matching the capabilities of theoretically-grounded algorithms with the practical demands of real-world applications. I have applied my research in domains such as humanoid soccer robots and rescue robots. I actively participate and contribute to events such as Robocup. My research is summarized by the following projects and videos:
Stochastic Search For Motor Learning
Stochastic search algorithms are general black-box optimizers. Due to their ease of use and their generality, they have recently also gained a lot of attention in operations research, machine learning and policy search. Yet, these algorithms require a lot of evaluations of the objective, scale poorly with the problem dimension, are affected by highly noisy objective functions and may converge prematurely. To alleviate these problems, we introduce a new surrogate-based stochastic search approach. We learn simple, quadratic surrogate models of the objective function. As the quality of such a quadratic approximation is limited, we do not greedily exploit the learned models. The algorithm can be misled by an inaccurate optimum introduced by the surrogate. Instead, we use information theoretic constraints to bound the `distance’ between the new and old data distribution while maximizing the objective function. Additionally the new method is able to sustain the exploration of the search distribution to avoid premature convergence. A video of this project is available at: [ https://www.youtube.com/watch?v=DB93_ojldKo] . In this video we used our method along with imitation learning to learn a table tennis stroke.
Contextual Stochastic Search For Multi Task Learning
Many stochastic search algorithms require relearning if the task changes slightly to adapt the solution to the new situation or the new context. In this project we want to find good parameter vectors for multiple related tasks, where each task is described by a continuous context vector. Hence, the objective might change slightly for each parameter vector evaluation. Therefore we we investigated the contextual stochastic search algorithms that can learn from multiple tasks simultaneously. Two videos of this project are available at [ https://www.youtube.com/watch?v=eMTr4hA37Dw ] and at [ https://www.youtube.com/watch?v=bKNBMPIEWQw ].
Humanoid Robot Omnidirectional Walking
In this project, we propose a novel omnidirectional walking engine that achieves human like, stable and fast walking. We augment the 3D inverted pendulum with a spring model to implement a height change in the robot’s center of mass trajectory. This model is used as simplified model of the robot and the zero moment point (ZMP) criterion is used as the stability indicator. Two videos of the project are available at [ https://www.youtube.com/watch?v=jHTba15b01M ] and at [ https://www.youtube.com/watch?v=z_0Udxk0VUg ].
Robocup Humanoid Soccer Simulation
In this project we build a team of simulated humanoid robots in context of FCPortugal project. We actively participate in Robocup competitions and we have won several titles. A video of this project is available at: [ https://www.youtube.com/watch?v=EwYBBUOF6Ao ]
Intelligent Wheelchair Simulator
In this project we developed a virtual intelligent wheelchair with multimodal interface. A video of this project is available at: [ https://www.youtube.com/watch?v=1enZjUB90eU ]