Hiroshi Tsujino
Hiroshi Tsujino has a total of 121 co-authors. Here are the top co-authors by number of shared publications:
Kazuhiro Nakadai
Honda Research Institute Japan Co., Ltd., Saitama, Japan
22 shared publications- Hiroshi G. Okuno
Waseda University,Tokyo,Japan; Graduate School of Informatics,Kyoto University,Kyoto,Japan, Japan
21 shared publications - Mikio Nakano
Honda Research Institute Japan Co., Ltd
18 shared publications Yuji Hasegawa
Honda Research Institute Japan Co., Ltd., 8-1 Honcho, Wako, Saitama 351-0188, JAPAN
18 shared publications- Kazuhiro Nakadai
Institute of Science Tokyo
12 shared publications Hirofumi Nakajima
Honda Research Institute Japan Co., Ltd, 8-1 Honcho, Wako-shi, Saitama 351-0114, Japan
8 shared publications
Latest Publications
- 15 April 2013
- Published by Taylor & Francis in Advanced Robotics
- Vol. 27 (10) , 749-758
Reinforcement learning system based on heuristics free state focusing
We discuss the role of state focus in reinforcement learning (RL) systems that are applicable to mechanical systems including robots. Although the concept of the state focus is similar to attention/focusing in visual domains, its implementation requires some theoretical background based on RL. We propose an RL system that effectively learns how to choose the focus simultaneously with how to achieve a task. This RL system does not need heuristics for the adaptation of its focus. We conducted a capture experiment to compare the learning speed between the proposed system and the traditional systems, SARSAs, and conducted a navigation experiment to confirm the applicability of the proposed system to a realistic task. In the capture experiment, the proposed system learned faster than SARSAs. We visualized the developmental process of the focusing strategy in the proposed system using a Q-value analysis technique. In the navigation task, the proposed system demonstrated faster learning than SARSAs in the realistic task. The proposed system is applicable to a wide class of RLs that are applicable to mechanical systems including robots.
- 15 April 2013
- Published by Taylor & Francis in Advanced Robotics
- Vol. 27 (10) , 749-758
Reinforcement learning system based on heuristics free state focusing
We discuss the role of state focus in reinforcement learning (RL) systems that are applicable to mechanical systems including robots. Although the concept of the state focus is similar to attention/focusing in visual domains, its implementation requires some theoretical background based on RL. We propose an RL system that effectively learns how to choose the focus simultaneously with how to achieve a task. This RL system does not need heuristics for the adaptation of its focus. We conducted a capture experiment to compare the learning speed between the proposed system and the traditional systems, SARSAs, and conducted a navigation experiment to confirm the applicability of the proposed system to a realistic task. In the capture experiment, the proposed system learned faster than SARSAs. We visualized the developmental process of the focusing strategy in the proposed system using a Q-value analysis technique. In the navigation task, the proposed system demonstrated faster learning than SARSAs in the realistic task. The proposed system is applicable to a wide class of RLs that are applicable to mechanical systems including robots.
- 1 July 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 173-179
Motion planning using state-dispersion-based phase space partition
This paper addresses a phase space partitioning problem in motion planning systems. In a previous study, we developed a kinematic and dynamic motion planning system, known as rapid semi-optimal motion-planning (RASMO), that ensures the optimality of the planned motions with rapid calculations using a partition for the phase space. The shape of the partition determines the optimality of the motion. We propose a state-dispersion-based phase space partitioning (SDPP) method that generates adaptive partitions for RASMO and the same class of motion planning systems. These partitions allow motion planning systems to plan motions with better optimality. To validate SDPP method, we compared the optimality of RASMO in several conditions using a double inverted pendulum model while setting the optimality criterion of RASMO to time. Results show that RASMO with SDPP planned smaller time motions than that obtained RASMO with a uniform partition. Once this method is applied to a machine (e.g. industrial or space robots), the planning system provides better motions with the same calculation cost.
- 1 July 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 173-179
Motion planning using state-dispersion-based phase space partition
This paper addresses a phase space partitioning problem in motion planning systems. In a previous study, we developed a kinematic and dynamic motion planning system, known as rapid semi-optimal motion-planning (RASMO), that ensures the optimality of the planned motions with rapid calculations using a partition for the phase space. The shape of the partition determines the optimality of the motion. We propose a state-dispersion-based phase space partitioning (SDPP) method that generates adaptive partitions for RASMO and the same class of motion planning systems. These partitions allow motion planning systems to plan motions with better optimality. To validate SDPP method, we compared the optimality of RASMO in several conditions using a double inverted pendulum model while setting the optimality criterion of RASMO to time. Results show that RASMO with SDPP planned smaller time motions than that obtained RASMO with a uniform partition. Once this method is applied to a machine (e.g. industrial or space robots), the planning system provides better motions with the same calculation cost.
- Open Access
- 2 March 2012
- Published by Springer Nature in EURASIP Journal on Audio, Speech, and Music Processing
- Vol. 2012 (1) , 12
A role of multi-modal rhythms in physical interaction and cooperation
As fundamental research for human-robot interaction, this paper addresses the rhythmic reference of a human while turning a rope with another human. We hypothyzed that when interpreting rhythm cues to make a rhythm reference, humans will use auditory and force rhythms more than visual ones. We examined 21-23 years old test subjects. We masked perception of each test subject using 3 kinds of masks, an eye-mask, headphones, and a force mask. The force mask is composed of a robot arm and a remote controller. These instruments allow a test subject to turn a rope without feeling force from the rope. In the first experiment, each test subject interacted with an operator that turned a rope with a constant rhythm. 8 experiments were conducted for each test subject that wore combinations of masks. We measured the angular velocity of force between a test subject/the operator and a rope. We calculated error between the angular velocities of the force directions, and validated the error. In the second experiment, two test subjects interacted with each other. 1.6 - 2.4 Hz auditory rhythm was presented from headphones so as to inform target turning frequency. Addition to the auditory rhythm, the test subjects wore eye-masks. The first experiment showed that visual rhythm has little influence on rope-turning cooperation between humans. The second experiment provided firmer evidence for the same hypothesis because humans neglected their visual rhythms.
- 2 March 2012
- Published by Springer Nature in EURASIP Journal on Audio, Speech, and Music Processing
- Vol. 2012 (1) , 12
Open AccessA role of multi-modal rhythms in physical interaction and cooperation
As fundamental research for human-robot interaction, this paper addresses the rhythmic reference of a human while turning a rope with another human. We hypothyzed that when interpreting rhythm cues to make a rhythm reference, humans will use auditory and force rhythms more than visual ones. We examined 21-23 years old test subjects. We masked perception of each test subject using 3 kinds of masks, an eye-mask, headphones, and a force mask. The force mask is composed of a robot arm and a remote controller. These instruments allow a test subject to turn a rope without feeling force from the rope. In the first experiment, each test subject interacted with an operator that turned a rope with a constant rhythm. 8 experiments were conducted for each test subject that wore combinations of masks. We measured the angular velocity of force between a test subject/the operator and a rope. We calculated error between the angular velocities of the force directions, and validated the error. In the second experiment, two test subjects interacted with each other. 1.6 - 2.4 Hz auditory rhythm was presented from headphones so as to inform target turning frequency. Addition to the auditory rhythm, the test subjects wore eye-masks. The first experiment showed that visual rhythm has little influence on rope-turning cooperation between humans. The second experiment provided firmer evidence for the same hypothesis because humans neglected their visual rhythms.
- 1 November 2011
- Published by Elsevier in Neural Networks
- Vol. 24 (9) , 950-960
Real-time simulation of a spiking neural network model of the basal ganglia circuitry using general purpose computing on graphics processing units
- 1 November 2011
- Published by Elsevier in Neural Networks
- Vol. 24 (9) , 950-960
Real-time simulation of a spiking neural network model of the basal ganglia circuitry using general purpose computing on graphics processing units
- 1 July 2011
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 766-771
Rapid approximation for optimal paths in phase space
This paper addresses optimal motion for general machines. Approximation for optimal motion needs a global path planning algorithm that precisely calculates the whole dynamics of a machine in a brief calculation. We propose a path planning algorithm that is composed of a path searching algorithm and a pruning algorithm. The pruning algorithm is based on our analysis for the resemblances of states. To confirm the precision, calculation cost, optimality, and applicability of the proposed algorithm, we conducted several shortest time path planning examinations for the dynamic models of double inverted pendulums. The precision to reach the goal state of the pendulums was better than other algorithms. The calculation was at least 58 times faster. There was a positive correlation between the optimality and the resolutions of the proposed algorithm. As a result of torque based feedback control simulation, we confirmed applicability of the proposed algorithm under noisy situation.
- 1 July 2011
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 766-771
Rapid approximation for optimal paths in phase space
This paper addresses optimal motion for general machines. Approximation for optimal motion needs a global path planning algorithm that precisely calculates the whole dynamics of a machine in a brief calculation. We propose a path planning algorithm that is composed of a path searching algorithm and a pruning algorithm. The pruning algorithm is based on our analysis for the resemblances of states. To confirm the precision, calculation cost, optimality, and applicability of the proposed algorithm, we conducted several shortest time path planning examinations for the dynamic models of double inverted pendulums. The precision to reach the goal state of the pendulums was better than other algorithms. The calculation was at least 58 times faster. There was a positive correlation between the optimality and the resolutions of the proposed algorithm. As a result of torque based feedback control simulation, we confirmed applicability of the proposed algorithm under noisy situation.
- 29 March 2011
- Published by Springer Nature in Applied Intelligence
- Vol. 34 (3) , 360-371
Ego noise cancellation of a robot using missing feature masks
- 29 March 2011
- Published by Springer Nature in Applied Intelligence
- Vol. 34 (3) , 360-371
Ego noise cancellation of a robot using missing feature masks
- 1 March 2011
- Published by Elsevier in Knowledge-Based Systems
- Vol. 24 (2) , 248-256
A multi-expert model for dialogue and behavior control of conversational robots and agents
- 1 March 2011
- Published by Elsevier in Knowledge-Based Systems
- Vol. 24 (2) , 248-256
A multi-expert model for dialogue and behavior control of conversational robots and agents
- 1 January 2011
- Published by Taylor & Francis in Advanced Robotics
- Vol. 25 (3-4) , 491-506
Physical Control of the Rotation of a Flexible Object — Rope Turning with a Humanoid Robot
Rope turning tasks are useful to explore rhythmic physical human–robot interaction. However, in traditional studies, a robot was not able to turn a rope by itself, because simultaneous control of three factors, i. e., energy transmission, rotational axis and centrifugal force, is difficult when a robot rotates a flexible object such as a rope. In this paper, we propose a method to control these three factors simultaneously. We developed the method by adding a compensator to an attractor that attracts the end-effector of a robot to a uniform circular motion within a fixed radius. In a rope turning simulation, the end-effector of a robot and the center of mass of a simplified rope converged to uniform circular motions. In addition, we applied the proposed method to a rope turning task performed by a humanoid robot. The robot was able to turn a rope with one fixed end or in cooperation with a human.
- 1 January 2011
- Published by Taylor & Francis in Advanced Robotics
- Vol. 25 (3-4) , 491-506
Physical Control of the Rotation of a Flexible Object — Rope Turning with a Humanoid Robot
Rope turning tasks are useful to explore rhythmic physical human–robot interaction. However, in traditional studies, a robot was not able to turn a rope by itself, because simultaneous control of three factors, i. e., energy transmission, rotational axis and centrifugal force, is difficult when a robot rotates a flexible object such as a rope. In this paper, we propose a method to control these three factors simultaneously. We developed the method by adding a compensator to an attractor that attracts the end-effector of a robot to a uniform circular motion within a fixed radius. In a rope turning simulation, the end-effector of a robot and the center of mass of a simplified rope converged to uniform circular motions. In addition, we applied the proposed method to a rope turning task performed by a humanoid robot. The robot was able to turn a rope with one fixed end or in cooperation with a human.
- 1 January 2011
- Published by Taylor & Francis in Advanced Robotics
- Vol. 25 (11-12) , 1405-1426
Whole Body Motion Noise Cancellation of a Robot for Improved Automatic Speech Recognition
The motors of a robot produce ego-motion noise that degrades the quality of recorded sounds. This paper describes an architecture that enhances the capability of a robot to perform automatic speech recognition (ASR) even as the entire body of the robot moves. The architecture consists of three blocks: (i) a multichannel noise reduction block, consisting of microphone-array-based sound localization, geometric source separation and post-filtering, (ii) a single-channel template subtraction block and (iii) an ASR block. As the first step of our analysis strategy, we divided the whole-body motion noise problem into three subdomains of arm, leg and head motion noise, according to their intensity levels and spatial location. Subsequently, by following a synthesis-by-analysis approach, we determined the best method for suppressing each type of ego-motion noise. Finally, we proposed to utilize a control module in our ASR framework; this module was designed to make decisions based on instantaneously detected motions, allowing it to switch to the most appropriate method for the current type of noise. This proposed system resulted in improvements of up to 50 points in word correct rates compared with results obtained by single microphone recognition of arm, leg and head motions.
- 1 January 2011
- Published by Taylor & Francis in Advanced Robotics
- Vol. 25 (11-12) , 1405-1426
Whole Body Motion Noise Cancellation of a Robot for Improved Automatic Speech Recognition
The motors of a robot produce ego-motion noise that degrades the quality of recorded sounds. This paper describes an architecture that enhances the capability of a robot to perform automatic speech recognition (ASR) even as the entire body of the robot moves. The architecture consists of three blocks: (i) a multichannel noise reduction block, consisting of microphone-array-based sound localization, geometric source separation and post-filtering, (ii) a single-channel template subtraction block and (iii) an ASR block. As the first step of our analysis strategy, we divided the whole-body motion noise problem into three subdomains of arm, leg and head motion noise, according to their intensity levels and spatial location. Subsequently, by following a synthesis-by-analysis approach, we determined the best method for suppressing each type of ego-motion noise. Finally, we proposed to utilize a control module in our ASR framework; this module was designed to make decisions based on instantaneously detected motions, allowing it to switch to the most appropriate method for the current type of noise. This proposed system resulted in improvements of up to 50 points in word correct rates compared with results obtained by single microphone recognition of arm, leg and head motions.
- Open Access
- 1 January 2011
- Published by The Robotics Society of Japan in Journal of the Robotics Society of Japan
- Vol. 27 (7) , 793-801
Musical Beat-Tracking for Robots and Its Application to A Music Robot
- 1 January 2011
- Published by The Robotics Society of Japan in Journal of the Robotics Society of Japan
- Vol. 27 (7) , 793-801
Open AccessMusical Beat-Tracking for Robots and Its Application to A Music Robot