Six-Legged Robot Snowball

General

The six-legged robot SNOWBALL was developed in 2004 with educational purposes in mind. It is meant to replace the rather rickety RugWalker in the RT-II-laboratory. Apart from that, it is used for project works that allow students to develop and test their own hard- and software extensions. One crucial design parameter for this robot was, that the components (apart from the microcontroller) should be comparably cheap standard parts.

System description

The acronym SNOWBALL stands for "System (that can) Not Only Walk But Also roLL" and indicates the hybrid approach to locomotion. It is well known that wheel-based locomotion is very energy efficient and fast on even surfaces. On the other hand it is as well known that in uneven or structured surroundings wheels reach their limits and legs are preferable. Most mobile robots nevertheless use only one of these two possibilities. There are some hybrid approaches, however. But most of the hybrid robots feature a rather complicated locomotion apparatus that would not comply with our attempt to keep the robot simple and cheap. The solution was to build a legged robot that is equipped with additional, passive castors. This enables the robot to walk over rough terrain and, when it reaches even ground, to lower on the castors and push forward with an arbitrary number of legs.

The legs of the robot are designed in a way to concentrate the mass in the torso. This means that all motors are located inside the torso together with accumulators and computational/steering units.

As can be seen in the pictures, SNOWBALL has twelve degrees of freedom, that is two per leg. With these two DOF it is possible to set the angle of upper and lower leg independently (within certain limits). By means of the inverse kinematics implemented in SNOWBALL's software structure, it is also possible to directly program the foot trajectories in a Cartesian coordinate system x-y that lies in the sagittal plane of the robot. To make the robot walk curves or to turn in place the right and left legs are move with different respectively opposing step widths. A variety of statically or dynamically stable trajectories is possible, but so far only the classic tripod trajectories has been implemented.

The robot is controlled by a Freescale MPC555 microcontroller that generates the required PWM-signals for the motors and provides enough I/O-pins for further applications.

Currently there are five ultrasonic distance sensors with a range of up to 8 m on the robot. Furthermore, an optical dead reckoning system has been developed during a project work. This system is based on two sensors taken from optical mice that detect movements over ground. As these sensors are very sensitive to variations of the distance to the ground, they can (as yet) only be used for motion measurement in the robot's rolling mode. At present the mice sensors are not attached to the robot but to an evaluation model.


	Six-Legged Robot Snowball General The six-legged robot SNOWBALL was developed in 2004 with educational purposes in mind. It is meant to replace the rather rickety RugWalker in the RT-II-laboratory. Apart from that, it is used for project works that allow students to develop and test their own hard- and software extensions. One crucial design parameter for this robot was, that the components (apart from the microcontroller) should be comparably cheap standard parts. System description The acronym SNOWBALL stands for "System (that can) Not Only Walk But Also roLL" and indicates the hybrid approach to locomotion. It is well known that wheel-based locomotion is very energy efficient and fast on even surfaces. On the other hand it is as well known that in uneven or structured surroundings wheels reach their limits and legs are preferable. Most mobile robots nevertheless use only one of these two possibilities. There are some hybrid approaches, however. But most of the hybrid robots feature a rather complicated locomotion apparatus that would not comply with our attempt to keep the robot simple and cheap. The solution was to build a legged robot that is equipped with additional, passive castors. This enables the robot to walk over rough terrain and, when it reaches even ground, to lower on the castors and push forward with an arbitrary number of legs. The legs of the robot are designed in a way to concentrate the mass in the torso. This means that all motors are located inside the torso together with accumulators and computational/steering units. As can be seen in the pictures, SNOWBALL has twelve degrees of freedom, that is two per leg. With these two DOF it is possible to set the angle of upper and lower leg independently (within certain limits). By means of the inverse kinematics implemented in SNOWBALL's software structure, it is also possible to directly program the foot trajectories in a Cartesian coordinate system x-y that lies in the sagittal plane of the robot. To make the robot walk curves or to turn in place the right and left legs are move with different respectively opposing step widths. A variety of statically or dynamically stable trajectories is possible, but so far only the classic tripod trajectories has been implemented. The robot is controlled by a Freescale MPC555 microcontroller that generates the required PWM-signals for the motors and provides enough I/O-pins for further applications. Currently there are five ultrasonic distance sensors with a range of up to 8 m on the robot. Furthermore, an optical dead reckoning system has been developed during a project work. This system is based on two sensors taken from optical mice that detect movements over ground. As these sensors are very sensitive to variations of the distance to the ground, they can (as yet) only be used for motion measurement in the robot's rolling mode. At present the mice sensors are not attached to the robot but to an evaluation model.

	Institute of Automatic Control - URL http://www.irt.uni-hannover.de/forschung/asr/six_en.html Responsible: Institute of Automatic Control, Biped Team, last modification 01.09.2009 © Leibniz Universität Hannover 1998 - 2009