arXiv subjects

Shinichi Kikuchi

Publications and source records attributed to Shinichi Kikuchi.

2 recordsLinked to original sources

arXiv2020

Technology to Counter Online Flaming Based on the Frequency-Dependent Damping Coefficient in the Oscillation Model

Online social networks, which are remarkably active, often experience explosive user dynamics such as online flaming, which can significantly impact the real world. However, countermeasures based on social analyses of the individuals causing flaming are too slow to be effective because of the rapidity with which the influence of online user dynamics propagates. A countermeasure technology for the flaming phenomena based on the oscillation model, which describes online user dynamics, has been proposed; it is an immediate solution as it does not depend on social analyses of individuals. Conventional countermeasures based on the oscillation model assume that the damping coefficient is a constant regardless of the eigenfrequency. This assumption is, however, problematic as the damping coefficients are, in general, inherently frequency-dependent; the theory underlying the dependence is being elucidated. This paper discusses a design method that uses the damping coefficient to prevent flaming under general conditions considering the frequency-dependence of the damping coefficient and proposes a countermeasure technology for the flaming phenomena.

cs.SI↗

arXiv2020

Learning Force Control for Contact-rich Manipulation Tasks with Rigid Position-controlled Robots

Reinforcement Learning (RL) methods have been proven successful in solving manipulation tasks autonomously. However, RL is still not widely adopted on real robotic systems because working with real hardware entails additional challenges, especially when using rigid position-controlled manipulators. These challenges include the need for a robust controller to avoid undesired behavior, that risk damaging the robot and its environment, and constant supervision from a human operator. The main contributions of this work are, first, we proposed a learning-based force control framework combining RL techniques with traditional force control. Within said control scheme, we implemented two different conventional approaches to achieve force control with position-controlled robots; one is a modified parallel position/force control, and the other is an admittance control. Secondly, we empirically study both control schemes when used as the action space of the RL agent. Thirdly, we developed a fail-safe mechanism for safely training an RL agent on manipulation tasks using a real rigid robot manipulator. The proposed methods are validated on simulation and a real robot, an UR3 e-series robotic arm.

cs.LG↗