RESEARCH VIDEOS

Robot Navigation

• RG on a long patrol in the USU Center for Persons with Disabilities (July 2004).
• Navigating in and out of an elevator (June 2004). Elevators are very difficult to handle for robotic guides. Human-robot interaction becomes very important. In this video, RG navigates in and out of an elevator.
• Crossing an open space at the USU CS Department (May 2004). Open spaces are a major challenge to mobile robots operating in indoor environments. In this video, RG is crossing an small open space at the USU CS Department.
• Here are three videos that compare sonar-based and laser-based navigation on two Pioneer 2DX bases in narrow and wide hallways. Same algorithms, same platforms, different sensors. The differences are remarkable.
  • Sonar-based navigation in a narrow hall. (April 2004)
  • Laser-based navigation in a narrow hall. (April 2004)
  • Both types of navigation in a wide hall. (April 2004)
• RG's obstacle avoidance (Part I) (Spring 2004). . RG avoids obstacles in a hallway at the USU CS Department.
• RG's obstacle avoidance (Part I) (Spring 2004). . RG avoids obstacles in a hallway at the USU CS Department.
• RG's obstacle avoidance (Part II) (Spring 2004). . RG avoids obstacles in a hallway at the USU CS Department.
• RG handling a hallway blockage (Part I) (Spring 2004). . How RG handles a situation when a hallway is blocked.
• RG handling a hallway blockage (Part II) (Spring 2004). . How RG handles a situation when a hallway is blocked.
• Sometimes RG misses RFID tags because it has to avoid obstacles, e.g., people sitting or standing near the tags. The two clips below show how this happens.
  • RG does not miss a tag (Spring 2004). This clip shows how RG detects the tag that triggers a correct left-turn behavior.
  • RG misses a tag due to obstacle avoidance( Spring 2004). This is the same root as in the above cliop. This time, however, RG misses the left-turn tag because it has to avoid a person sitting under the tag.
• RG navigates an area at the USU CS Department alone (Winter 2004). The video shows the integration of the laser range finder and RFID. The white masking tapes on the floor were used in our experiments to estimate the smoothness of RG's navigation. They do not carry any information for the robot, because the robot cannot sense them.
• Robot Alone (Summer 2003). RG navigates an area at the USU Assistive Technology Lab. Each destination is coded as a number which RG receives through a keypad. Sometimes RG's sonars crosstalk which causes the robot to make jerky movements and change its orientation. Sometimes such jerky movements make the robot face a wall. The recovery strategy is quite simple: assume that it is an obstacle and ask it to yield the way. If the obstacle persists, either avoid it or allign with it and then follow it because it must have been a wall.
• Making a U-Turn (Spring 2003). RG makes a u-turn in a hallway.
• Crowded Environment (Spring 2003). RG operates in a crowded environment.

Human-Robot Interaction

• Robot-assisted navigation experiments at the USU CS Department (Part III) (Winter 2004). A closer look at the human-robot interaction through a wearable keypad. Sensor fusion of laser range finding and RFID.
• Robot-Assisted Navigation with Speech-based Interaction (Summer 2003). The video shows how RG helps a visually impaired person to navigate to a student's office, the rest room, and the main office of the USU CS Department. All interaction is speech-based.
• Asking a Person to Yield (Summer 2003) . When an obstacle is detected in front of the robot, RG assumes that it is a person and asks the person to yield the way.
• Keypad Training (Spring 2003). In many environments and with many people, speech is error-prone. An alternative is to use a keypad to communicate with the robotic guide. This video show a brief keypad training received by a subject before she uses the robot to navigate to various locations.
• Quick Human-Robot Interaction Training (Spring 2003). Visually impaired people who have never used guide dogs need a brief training on the basic concepts of following an external guide. This video shows how a visually impaired person who has never used a guide dog before receives a brief introduction to RG before he is asked to use RG to navigate the CS Department's office space.
• These videos show how our visually impaired subjects interact with RG through a keypad.
  • Keypad-Based Interaction (Part I) (Spring 2003). A subject navigates part of the office space of the CS Department using a keypad to communicate with the robot..
  • Keypad-Based Interaction (Part II) (Spring 2003). A subject navigates another route in the office space of the CS Department using a keypad to communicate with the robot..
  • Keypad-Based Interaction (Part III) (Spring 2003). RG guides a visually impaired person at the USU Assistive Technology Lab. The keypad is used to communicate with the robot.
  • Keypad-Based Interaction (Part IV) (Spring 2003). RG guides a visually impaired person in the CS Department's office space. The keypad is used to communicate with the robot.
  • Keypad-Based Interaction (Part V) (Spring 2003). RG guides a visually impaired person in the CS Department's office space. The keypad is used to communicate with the robot.
• Other robot videos. Voice interaction with a mobile robot.

Wearable Wayfinding Toolkit for Guide Dog Users

A Wearable Wayfinding Toolkit (WWT), our most recent project, came out as a result of our research on sensor fusion in RG. The WWT fuses three sensors: a GPS, an orientation sensor, and a wireless LAN sensor. It is designed to be worn by a guide dog user in unfamiliar environments. It does localization, tracking, and orientation. Immediate obstacle avoidance is delegated completely to the dog. The videos below show the WWT and how it can be worn by the user.

• Main Quad of USU Campus (June 2004). A short clip showing the Main Quad of USU Campus where we conduct outdoor WWT-based navigation experiments.
• WWT explained (June 2004). John Nicholson, a CS Ph.D. student, explains what the WWT and how it is put together.
• Putting the WWT on (June 2004). Sachin Pavithran, a visually impaired member of our research team and a guide dog user, is trying the WWT on.
• Putting the WWT on (June 2004). Sachin Pavithran, a visually impaired member of our research team and a guide dog user, is trying the WWT on.
• WWT-Based Navigation on the Main Quad of USU Campus (June 2004). Sachin Pavithran, a visually impaired member of our research team and a guide dog user, uses the WWT to navigate simple routes on the Main Quad of USU Campus.
• WWT-Based navigation and obstacle avoidance (June 2004).. This video features how Sachin Pavithran's guide dog, Barstow, avoids a large vehicle parked on a side walk on the Main Quad of the USU Campus (June 2004).

Exit Interviews

• What do some visually impaired participants we have worked with think of our robot-assisted navigation? Here are a few exit interviews. You may want to view these after viewing the videos on robot-assisted navigation and human-robot interaction above.
  • Exit interview 1 (Summer 2003).
  • Exit interview 2 (Summer 2003).
  • Exit interview 3 (Summer 2003).
  • Exit interview 4 (Summer 2003).
  • Exit interview 5 (Summer 2003).