Autonomous mobile robotic manipulation
Mobile manipulation, the subspecialty of robotics concerned with the close coupling of navigation and manipulation, has drastically grown in popularity in recent years. This has largely been driven by decreasing hardware costs, particularly in sensors and actuators, and the proliferation of the online open-source software community.
The effect is a wide and increasing range of robotics applications, where robots interact with their surroundings without being fixed to a single location. It is through these capabilities that robots may finally realise their full potential in domains such as healthcare and rehabilitation, search and rescue, as well as assisted living.
Over the last few years, several robotics manufacturers have responded to the changing state of research, software and hardware and have commercialised mobile manipulators. Although complete off-the-shelf mobile manipulation systems are available, assembling a mobile manipulator in-house provides the benefit of increased flexibility in platform design. This also results in the reuse of existing hardware with the effect of considerably lower costs compared with purchasing new equipment. Furthermore, there is a range of existing designs varying in size, capabilities and constituent parts from which to draw inspiration.
CSIR researchers assembled an entire hybrid autonomous manipulation platform in-house to be used for research into autonomous mobile manipulation. It includes a robotic arm (Barrett Whole Arm Manipulator) with seven degrees of freedom and a sturdy four-wheeled base (two active and two passive wheels) with differential drive (the Powerbot AGV). The robot is equipped with an additional computer for control and sensory processing, as well as several sensors.
The combination of these two off-the-shelf robots, with the incorporation of range- and depth-sensing capabilities, has resulted in a fully-functional multi-purpose system, which is wholly compatible with a community of open-source software and suited to a diverse range of applications.
The integrated platform has a reach of about 1 m, from a shoulder height of about 0.83 m, allowing it to manipulate objects placed on standard desks and tables, as well as reach door handles, elevator buttons and so forth. It has a top speed of 6 km/h and the arm has a three-fingered hand, which can lift a payload of 2 kg. The entire system has a battery life of approximately two to three hours.
While the aim of developing this platform is for it to be used in a variety of tasks under autonomous operation, it has also been configured for manual control. This is done by means of joystick teleoperation from an external operator's console that connects wirelessly to the platform.
Researchers continue their work to increase the autonomy, robustness and usability of the platform.
Dr Benjamin Rosman