Landing control of acrobat robot (SMB) satisfying various constraints

Sadahiro, Teruyoshi; Yamakita, Masaki

doi:10.1109/robot.2003.1242054

Cited by 5 publications

(2 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A common application involves solving physical contact problems [1,15,17,20,30,31,33]. LCPs have also been applied to the study of elastoplastic structures and nonlinear obstacle problems [15,26], the analysis of a step down (buck) dc-dc power converter [3], and the study of finance models [26].…”

Section: Ls Backgroundmentioning

confidence: 99%

Distributed identification of the lineality space of a cone

2008

View full text Add to dashboard Cite

A distributed approach is described for solving lineality (or linearity) space (LS) problems with large cardinalities and a large number of dimensions. The LS solution has applications in engineering, science, and business, and includes a subset of solutions of the more general extended linear complementarity problem (ELCP). A parallel MATLAB framework is employed and results are computed on an 8-node Rocks based cluster computer using Remote Procedure Calls (RPCs) and the MPICH2 Message Passing Interface (MPI). Results show that both approaches perform comparably when solving distributed LS problems. This indicates that when deciding which parallel approach to use, the implementation details particular to the method are the decisive factors, which in this investigation give MPICH2 MPI the advantage.

show abstract

Section: Ls Backgroundmentioning

confidence: 99%

Distributed identification of the lineality space of a cone

2008

View full text Add to dashboard Cite

show abstract

“…Specifically for the generation of acrobatic motion with a horizontal bar, Sadahiro realized taking off from the horizontal bar and somersault in the air [3], [4] and Kishikawa realized landing control on tip toe [5]. Though the landing control is realized using standing on tip toe, region of stabilized posture is not so large.…”

Section: Introductionmentioning

confidence: 99%

Falling avoidance control of acrobat robot by reinforcement learning

Kochiya

Yamakita

2007

SICE Annual Conference 2007

Self Cite

View full text Add to dashboard Cite

In this study a landing control of an acrobat robot is considered and Q-learning method is applied for falling avoidance control. Since the dynamics of the system is changed according to contact conditions to the ground, the system is a typical variable constraint and hybrid system. The state space for the Q-learning consists of discrete mode variable and continuous states. It is shown by numerical simulations that taking a step motion is automatically generated and falling down is avoided properly.

show abstract

Untitled

2002

View full text Add to dashboard Cite

Abstract. The problem of state abstraction is of central importance in optimal control, reinforcement learning and Markov decision processes. This paper studies the case of variable resolution state abstraction for continuous time and space, deterministic dynamic control problems in which near-optimal policies are required. We begin by defining a class of variable resolution policy and value function representations based on Kuhn triangulations embedded in a kd-trie. We then consider top-down approaches to choosing which cells to split in order to generate improved policies. The core of this paper is the introduction and evaluation of a wide variety of possible splitting criteria. We begin with local approaches based on value function and policy properties that use only features of individual cells in making split choices. Later, by introducing two new non-local measures, influence and variance, we derive splitting criteria that allow one cell to efficiently take into account its impact on other cells when deciding whether to split. Influence is an efficiently-calculable measure of the extent to which changes in some state effect the value function of some other states. Variance is an efficiently-calculable measure of how risky is some state in a Markov chain: a low variance state is one in which we would be very surprised if, during any one execution, the long-term reward attained from that state differed substantially from its expected value, given by the value function.The paper proceeds by graphically demonstrating the various approaches to splitting on the familiar, non-linear, non-minimum phase, and two dimensional problem of the "Car on the hill". It then evaluates the performance of a variety of splitting criteria on many benchmark problems, paying careful attention to their number-of-cells versus closeness-to-optimality tradeoff curves.

show abstract

Landing control of acrobat robot (SMB) satisfying various constraints

Cited by 5 publications

References 7 publications

Distributed identification of the lineality space of a cone

Distributed identification of the lineality space of a cone

Falling avoidance control of acrobat robot by reinforcement learning

Untitled

Contact Info

Product

Resources

About