Exploring the importance of embodiment in Assistive Robotics Author Name Lab Affiliation University of Southern California E-mail Address Motivation Humans impart a natural and social response to various forms of media and intelligent systems.1 Recent work in the area of human robot interaction has generated empirical results showing that different forms of media affect people differently and that there are benefits to physically embodied systems as opposed to other forms of interaction, such as television.2 Additional work has demonstrated that nationality and gender may play a role in the importance and effectiveness of embodiment.3,4 Continuing along those lines, there is a possibility that other factors such as health and age5 may also affect the effectiveness of embodiment. The elderly population in the United States is currently growing and technology posses potential to provide much needed care and support6 but the most effective forms of assistance must be determined. Goal The goal of this project will be to design and implement a test set-up which will be capable of generating empirical evidence on the efficiency and usefulness of embodiment in assistive robotics. Utilizing the previous work of the Interaction Laboratory's Post Cardiac Surgery Robot (named Clara), this project will focus on rehabilitation and recovery.7 Different test cases utilizing differing degrees of agent embodiment, by combining Clara with a handheld device, will be compared to reveal how cardiac patients respond to different stimuli and means of motivation. Description and Test bed Building on the Clara project, a wireless handheld device will be introduced to communicate with recovering patients. Utilizing a large touch sensitive display, the new device will allow Clara to receive patient input without the use of voice recognition. Additionally, the display will allow information and directions to be presented to the patients through non-auditory means. As many of the people targeted by assistive robotics may posses minimal technological experience, simplicity of use and clarity of action will be primary features. It is intended that this device be constructed from a Dell Axim X50 or X50v (see Figure 1 and Table 1). Since these devices posses buttons on the front of the enclosure, it is intended that a plastic or leather case will be used to cover the buttons. While useful for testing and interacting with technology accustomed humans, their removal from view may make the device less intimidating and easier to use for patients. To provide consistency with the Player/Gazebo software packages, Linux may (depending on compatibility with embedded hardware) replace the Windows operating system that comes standard with the device. Clara utilizes voice recognition via a microphone located on the mobile robot. Especially since hospital settings are the normal environment for Clara, the accuracy of this system is currently not at a desirable level of performance. Additionally, patients may not be capable of speaking at load volumes without incurring injury and will already be taxed from their breathing exercises. The new wireless device will be attached near by the patient and when voice recognition is used, the microphone (or headset jack) on the handheld device could be used to increase accuracy. This will be critical to ensure a fair comparison between different methodologies. Evaluation In order to test the effectiveness of embodiment, different scenarios will be constructed which differ in the degree of utilization of both Clara and the handheld device. Arranged in rough order of complexity, the intended scenarios are: Clara without the handheld device, introduction of the handheld device but only for touch based input, adding an information display to the handheld device, just the handheld device utilizing bidirectional communication, and finally, adding an animated face to the previous scenario. At a minimum, the first three scenarios will be constructed and if additional time is available, the remaining two will be realized.When the scenarios have been completed, testing will begin. Participants will not be actual cardiac patients (for safety and availability); instead USC students and faculty will be recruited. Each patient will need to go through each of the test set-ups. Different orderings will be used for each test subject. An evaluation will need to be made as to whether each participant should do all test setups at one time or over multiple testing periods (the amount of required set-up and change time will be a great factor). While it would be desired to have enough participants to make the testing statistically valid, this project is simply aiming for proof of principle. Additional testing can be executed during a second phase of this project. Figure 1: Dell Axim X50v Table 1: Dell Axim Relevant Parameters Dell X50 Dell X50v Processor Speed 520MHz 624MHz Screen Size 3.5 inches 3.7 inches Screen Resolution240 x 320 480 x 640 Graphics ? Intel 2700G w/ 16MBPrice $399.00 $499.00 Milestones 1. Acquire Dell Axim device and select appropriate software (Sept. 27 – Oct. 11) 2. Design user interfaces for handheld device (Oct 11 – Oct 25) 3. Design testing scenarios and locate/prepare test site ( Oct. 25 – Nov. 22) 4. Perform tests with test subjects to verify experimental design (Nov. 22 – Dec. 7) 5. Presentation (Dec. 7 – Dec. 8) Future Extensions Future extensions to this project can be separated into three areas. First, the results in this project may not directly correspond with the effectiveness of the test scenarios with actual cardiac patients due to differing demographics. Additional testing could provide more accurate numbers. Second, the task performed by Clara could be changed as patients with different illnesses may require different levels and modes of interaction. Additionally, the concept of co-learners8 could be converted into this domain with the handheld device being used to realize a software agent working with Clara. 1 B. Reeves, and C. Nass. The Media Equation: How People Treat Computers, Television, and New Media Like Real People and Places, Cambridge University Press, New York, NY, 1998. 2 C. Kidd, “Sociable Robots: The Role of Presence and Task in Human-Robot Interaction”, Diss, Massachusetts Institute of Technology, Jun 2003. 3 K. Shinozawa, B. Reeves, K. Wise,