Logo des Repositoriums

Auflistung nach Autor:in "Hamker, Fred"

1 - 2 von 2
  • Zeitschriftenartikel
    Spatial Cognition of Humans and Brain-inspired Artificial Agents
    (KI - Künstliche Intelligenz: Vol. 29, No. 1, 2015) Hamker, Fred
    Present vision systems primarily operate on still images or an image sequence but hardly consider continuous perception across actions. If sensors are attached to the body of a human-like agent who interacts with the environment, several questions arise about how to update the reference systems with each action. In our European research project “Spatial Cognition” we address this topic by a combination of experimental and computational work which should finally merge into a large-scale model of human-like space perception and spatial memory being tested on a humanoid agent in virtual reality.
  • Textdokument
    Training a deep policy gradient-based neural network with asynchronous learners on a simulated robotic problem
    (INFORMATIK 2017, 2017) Lötzsch, Winfried; Vitay, Julien; Hamker, Fred
    Recent advances in deep reinforcement learning methods have attracted a lot of attention, because of their ability to use raw signals such as video streams as inputs, instead of pre-processed state variables. However, the most popular methods (value-based methods, e.g. deep Q-networks) focus on discrete action spaces (e.g. the left/right buttons), while realistic robotic applications usually require a continuous action space (for example the joint space). Policy gradient methods, such as stochastic policy gradient or deep deterministic policy gradient, propose to overcome this problem by allowing continuous action spaces. Despite their promises, they suffer from long training times as they need huge numbers of interactions to converge. In this paper, we investigate in how far a recent asynchronously parallel actor-critic approach, initially proposed to speed up discrete RL algorithms, could be used for the continuous control of robotic arms. We demonstrate the capabilities of this end-to-end learning algorithm on a simulated 2 degrees-of-freedom robotic arm and discuss its applications to more realistic scenarios.