Artificial Hormone-based Architecture for Adaptable Robots
- Members: Pitiwut Teerakittikul, Gianluca Tempesti and Andy Tyrrell
- Start Date: January 2009
Downloads
- Hormone-inspired Controller Structure for Adaptable Robots, Poster as PDF.
Overview
Adaptability is an essential ability of autonomous robots working in real-world human-hazardous environments. These kinds of environments are dynamic and full of unexpected circumstances caused by environmental changes. In fact, changes can happen in both internal (e.g. faults occur on robot’s components or a reduction of robot’s power) and external systems (e.g. changes of terrain characteristics or roughness) of the robots.
Homeostasis is a crucial property in helping maintain internal states of an organism. Generally, in biological organisms, when environmental changes occur which disturb internal states of an organism, homeostasis plays a vital role in helping re-balance the internal states of an organism, and assist an organism to cope with the changes. It is understood that there are three main systems responsible to homeostasis which are the nervous, immune and endocrine systems. Basically, the endocrine system can be viewed as a system of glands which secrete hormones in response to changes of environments and also helping other functions of an organism.
The main purpose of this research is to utilize artificial hormones in order to provide adaptability for an autonomous 4-wheel robot in such a way that it can cope with both internal and external environmental changes.
Gazebo – 3D Robot Simulator
Our robot, environment and control models are implemented in Gazebo, an open-source 3D physically-realistic robot simulator. This simulator is developed under the Player and Stage projects (http://playerstage.sourceforge.net/). Gazebo is able to simulate both static and dynamic objects, such as robots or sensors and the rigid-body physics of robots, sensors, objects and interactions among them in simulated environments. Gazebo is a powerful tool for implementation of autonomous robot systems.
An example of a Gazebo environment
First Experiments
The main purpose of this experiment is to investigate the use of a hormone in helping a 4-wheel autonomous robot to negotiate a rough terrain which is an example of external environmental change.
Experimental setup
The autonomous robot is released in a rough terrain arena, as shown in the pictures above. The robot is expected to approach a targeted objected, in this case a green cylinder, while also negotiating rough terrain along the way.
A basic hormone responding to terrain roughness is used here on the robot. A greater degree of the terrain roughness causes the hormone concentration to increase, which then results in gradual decrease of the robot velocity.
First results
We have observed from first results that a number of robot’s tip-over rate is decreasing and the robot have shown a better performance in negotiating the terrain. Videos below show examples of robot behaviours when there is and there is no hormone implemented on the robot.
First video: when there is no hormone on the robot system
Second video: when there is hormone on the robot system