Designing Safety-Critical Rehabilitation Robots

In recent years, robots have made substantial in-roads in the medical field and are gradually finding their way into clinical practice. Intuitive Surgical’s da Vinci® surgical robot broke ground in 1998 by performing the first tele-robotic surgery to repair a heart valve (Salisbury, 1998; Guthart & Salisbury, 2000). Accuray’s CyberKnife radiotherapy robot began treating head, neck and upper spine tumors in 1999 by combining image guidance with a robotically-directed radiation beam (Adler et al., 1997). In 2002, Interactive Motion Technology began therapy of stroke patients with the InMotion2 robot, also known as the MIT-Manus (Krebs et al., 2002). These devices and many others under development have provided researchers and doctors alike with capabilities not previously available. These additional capabilities, however, have also brought with them the issue of safety – these are safety critical systems in which a single malfunction can endanger the life of the patient. In contrast with traditional robotic systems, medical robots must enforce the safety of the patient as an object within its workspace, while also being able to treat the patient. This dichotomy creates the need for a safety system that can allow the robot to interact with the patient, while also enforcing all necessary safety precautions. Human fatalities resulting from medical treatment with machines is unfortunately all-toreal. The Therac 25, a radiation therapy machine developed by the Atomic Energy Commission of Canada, was involved in six known accidents between 1984 and 1987. Five patients died as a result of massive overdoses of radiation when a high power electron beam was activated without the target tumor having been rotated into place (Leveson & Turner, 1993). Had the machine’s software detected the fault, the accident would have been averted. Radiation therapy machines are now required to have hardware interlocks to prevent activation of the high-energy electron-beam unless the target is in place. A similar tragedy occurred at the National Oncology Institute in Panama during 2000 and 2001. Twenty-eight patients were overexposed during radiation therapy for cancer treatment, after use of a computerized treatment planning system. Dosage calculations from the planning system had errors of up to 105%. By August 2005, 23 out of 28 overposed patients had died, of which at least 18 were attributed to radiation effects (Borras, 2006). Dangers in medical robotics are not confined to surgical systems. In powered orthoses or “exoskeletons” being developed for rehabilitation, humans are basically encapsulated in the device creating a potentially hazardous situation. Powered leg exoskeletons such as the the LokomatTM Gait Orthosis are being used to train stroke and spinal cord injury patients how to O pe n A cc es s D at ab as e w w w .ite ch on lin e. co m