Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.

  

A real-time optimal controller for the NAO robot locomotion system

Author(s):

D. Hug
Conference/Journal:

Semester Thesis, FS16
Abstract:

The walking engine, implementing the generation of motion for an arbitrary physical system, is ultimately responsible for giving appropriate commands to the systems physical servos. It is undoubtedly one of the core components that enable the device/robot to achieve all the possible tasks it was designed for. Especially in the case of a bipedal robot, it is crucial that the robot is able to move from a position A to a position B in a fast, stable and energy saving manner. Even more so, if said bipedal robot is intended to play football against other intelligently designed robots and relies on its agility and ability to achieve a certain strategic move to an overwhelming extent. It is evident that other very important components within the software of the robot take a big part in the overall performance as well. For instance, even a perfect working walking engine is insignificant in complete absence of an intelligent and well-performing cognitional module that defines the overall strategy to follow for the physical act of walking. Vice versa, even the sharpest artificial mind is worthless if its wisely chosen decision can not be realised by the physical layer of the system. This semester thesis was particularly concerned about the walking engine that is used within the NAO robot, which is used by the ETH NomadZ in the RoboCup tournament. The framework for the walking engine that was used so far is based on an extensive software framework that was released by B-Human, a competitor of the NomadZ, back in 2013. Although the framework shows an acceptable overall performance, there is, nonetheless, lot of room for substantial improvements in general. Specifically, in the case of the walking engine, the previously used algorithm exposes several disadvantages. One of the most eminent problems is that the algorithm gives rise to a generally unstable gait if employed in ordinary situations. Furthermore, the preceding algorithm recalculates an entirely new trajectory at a rate of approximately 50 [Hz]. Said updates are done in an trade-off based optimization that tries to satisfies certain properties (e.g. constant speed, selection of an appropriate pivot point etc.) when transitioning from one leg to the other. Though this method does have its advantages, computational cost is one of the key factors to consider on the real robot. This is due to the fact that the CPU on the actual NAO robot is quite weak compared to other contemporary CPUs. Lastly, so far, merely one semester thesis has been conducted in this direction. Namely, the work of S. Plüss with the title "Dynamic Analysis and Control of a Robotic Bipedal Locomotion System". This scarcity of expertise and dependency on an external software framework in the area of bipedal motion originally gave rise to start of projects in this area. One of the declared goals, within the course of several consecutive theses, is to develop a strong departmental knowledge in this domain. In the course of this thesis, a mathematical expression for the ideal trajectory of a bipedal walk has first been derived, then after, largely following the work of S. Plüss, the proposed trajectory/walking algorithm has been implemented and tested on the real robot as well as in SimRobot (simulation tool). In order to achieve this implementation, the old code/module has fundamentally been redeveloped and the problem has been broken down into following an ideal trajectory using a standard PID controller as a first, basic departmental implementation of a bipedal walk on the NAO.

Surpervisors: Darivianakis Georgios, Flamm Benjamin, John Lygeros

Year:

2017
Type of Publication:

(13)Semester/Bachelor Thesis
Supervisor:



File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2017:IFA_5619
}
Permanent link