AVI (DIVX)

640x480 (25 MB)

MOV (MPEG-4)

640x480 (22 MB)

320x240 (15 MB)

This clip shows a helicopter flight with a camcorder strapped to it. A SICK LMS laser was also strapped onto the helicopter during these flights. Although data was not collected from the laser future flights would involve data collection, autonomous 3-D mapping and nap of the earth flight.

AVI

640x480 (76 MB)

This clip shows a few successful runs of the heli turning away from obstacles on either side using the optic-flow based approach.

AVI

640x480 (82 MB)

320x240 (31 MB)

MOV (MPEG-4)

640x480 (60MB)

320x240 (27MB)

This clip shows a few successful runs of the heli turning away from a row of trees on its left hand side. The trees are sensed by comparing optic flow in the images from sideways-looking cameras on either side of the heli, and turning away from the side that has a greater flow.

AVI

640x480 (81 MB)

320x240 (30 MB)

MOV (MPEG-4)

640x480 (48 MB)

320x240 (23 MB)

Piloted data-logging flight at the Del Valle Urban Search and Rescue training site. The heli was piloted between various obstacles while images were logged, and this data will be used to test the optic-flow based obstacle avoidance technique

AVI

640x480 (87 MB)

320x240 (32 MB)

MOV (MPEG-4)

640x480 (125 MB)

320x240 (30 MB)

Autonomous deployment during autonomous flight. The heli flew autonomously to 40 GPS waypoints and deployed a marker at each point, forming a 5x8 grid at 2m intervals. Motes were placed where the markers fell, forming a network on the ground.

AVI

640x480 (116 MB)

320x240 (42 MB)

MOV (MPEG-4)

640x480 (154 MB)

320x240 (31 MB)

Autonomous Localization. After deploying the network of motes, the helicopter traversed the field a few times broadcasting GPS data to the motes, so they could localize themselves. This involved 4.5 minutes of autonomous flight while the heli passed through 10 GPS waypoints.

AVI

640x480 (53 MB)

320x240 (19 MB)

MOV (MPEG-4)

640x480 (60 MB)

320x240 (17 MB)

Autonomous Network Repair. The network of motes was split to create two separate networks. The heli was used to deploy more motes along the split in order to repair the network. This involved autonomous flight and deployment at a number of pre-determined GPS locations.

AVI

640x480 (78 MB)

320x240 (21 MB)

MOV (MPEG-4)

640x480 (77 MB)

320x240 (29 MB)

Camcorder Footage. This footage is taken from a camcorder which we mounted onboard AVATAR out of curiosity. The image stabilization function on the camera seems to be effective in dealing with the engine vibration. This was piloted flight.

AVI

640x480 (42 MB)

320x240 (12 MB)

MOV (MPEG-4)

640x480 (29 MB)

320x240 (10 MB)

Omnicam Footage. This footage was taken with the omnidirectional lens when we were still using an analog B&W ccd camera.

AVI

640x480 (58 MB)

320x240 (30 MB)

MOV (MPEG-4)

640x480 (36 MB)

320x240 (17 MB)

First stable autonomous flight with 3rd Generation AVATAR. During the first few seconds of this clip, the gains are being tuned, and then when the correct gains are used, the helicopter performs very stable autonomous flight.

AVI

640x480 (49 MB)

320x240 (20 MB)

MOV (MPEG-4)

640x480 (38 MB)

320x240 (18 MB)

Autonomous landing mishap. While testing autonomous landing on the latest AVATAR, we discovered a bug in the code that could have been disastrous if it were not for the quick reactions of our safety pilot Doug Wilson. Luckily, no damage was caused.

AVI
320x240 (25 MB)

MPG
160x120 (23 MB)

AVI
320x240 (15 MB)

MPG
160x120 (13 MB)

AVI
320x240 (17 MB)

MPG
160x120 (15 MB)

AVI
320x240 (14 MB)

MPG
160x120 (12 MB)

AVI
320x240 (18 MB)

MPG
160x120 (16 MB)

AVI
320x240 (13 MB)

MPG
160x120 (12 MB)

AVI
320x240 (11 MB)

MPG
160x120 (9 MB)

In this video, the AVATAR is flying autonomously while tracking an Urban Robot manufactured by IS Robotics.  The three perspectives in this video come from an external camera, the wireless video feed from the AVATAR, and screen captures of the computer running the Vision Tracking software (Mar-2000).

In this scenario, the helicopter flies above a "suspicious person" shown through the helicopter camera. When the person hides under a truck, the helicopter lands, and dismounts a small ground vehicle. The ground vehicle then spots the person under the truck with its own camera. The helicopter later returns to the ground, and reacquires the ground vehicle. (11-Dec-99). Both the helicopter and ground vehicle are manually operated.

DGPS-based navigation with semi-autonomous robot control (roll and pitch under robot control while yaw, throttle and collective under human control). The flight from left to right is under total human control while the return flight back to the starting goal location is with roll and pitch under robot control. The error in latitude and longitude is used to compute necessary lateral velocities to get back to the goal location.  These velocities in turn are used to generate desired roll and pitch angles to move the AVATAR laterally to the goal. (5-Dec-99)

This is a clip taken in September 1998, where the pilot is controlling the second generation AVATAR in an attempt to gather the flight data necessary to generate a fuzzy logic rule base for autonomous roll control. (Sep-98).

This is a video of the AVATAR crashing. It is this event that inspired James Montgomery to use a "teaching by showing," or learning approach for controller generation.

Here is the AFV in its 1st place performance in the 1994 Association for Unmanned Vehicle Systems, International's Aerial Robotics Competition.  The AFV is shown successfully flying over the disks that need to be picked up. (May-94).

Last modified: May 25 2006