Work bench and bot note:the elbow servo is missing
(took a call to hitec to find a replacement for
discontinued servo)
About a week ago we ordered a Lynxmotin (link) tracked chassis and AL5D arm and it came in the mail yesterday. Today we assembled the chassis, arm and bolted on the arm assembly. Fist impressions of the lexan framed track kit are of high quality and durablity, however it only has about and inch of ground clearance. This is fine for flat terrain but will present problems on mixed ground. The base of the arm is ABS plastic and the other components are aluminium brackets and fittings. We got the hardware only kit so as to upgrade to only metal geared servos.
The ultimate goal of this bot is to test and prototype passive sensing algorithms for navigation. Most robots that are aware of their surroundings do so by "active" sensing with either radar or sonar beams that are transmitted and reflected off of surfaces, or using beacon technology like GPS. These are reliable methods for orienteering but have limed range and are susceptible to interference and spoofing (read how Iran stole a spy drone, unconfirmed) There are plenty of inexpensive camera kits that include line sensing and object tracking on-board however the crux of the challenge is to interlace various frames to build a basic 3d environment using only a microprocessor architecture and an inertial sensor for hints. Case in point: imagine deploying and EOD bot during a complex attack. Once the bot has reached its goal and done its job, the last thing you want to do is drive it all the way back to where it started. The EOD tech would rather be on his rifle ready for a secondary attack - while the bot automatically 'comes home.' Or imagine a lightweight quad copter that automatically follows an unmounted squad and returns to the Humvee recharge. This would make battlefield robotics much easier to use and intuitive for the soldier to effectively use. In a civilian setting imagine a robotic assistant that followed you wherever you may go; no additional programming or beacon signals required.
Initially, the control electronics will be Arduino; we plan on switching to a beagle board once we get the hardware tested and the basics out of the way. Robot porn to follow:
The ultimate goal of this bot is to test and prototype passive sensing algorithms for navigation. Most robots that are aware of their surroundings do so by "active" sensing with either radar or sonar beams that are transmitted and reflected off of surfaces, or using beacon technology like GPS. These are reliable methods for orienteering but have limed range and are susceptible to interference and spoofing (read how Iran stole a spy drone, unconfirmed) There are plenty of inexpensive camera kits that include line sensing and object tracking on-board however the crux of the challenge is to interlace various frames to build a basic 3d environment using only a microprocessor architecture and an inertial sensor for hints. Case in point: imagine deploying and EOD bot during a complex attack. Once the bot has reached its goal and done its job, the last thing you want to do is drive it all the way back to where it started. The EOD tech would rather be on his rifle ready for a secondary attack - while the bot automatically 'comes home.' Or imagine a lightweight quad copter that automatically follows an unmounted squad and returns to the Humvee recharge. This would make battlefield robotics much easier to use and intuitive for the soldier to effectively use. In a civilian setting imagine a robotic assistant that followed you wherever you may go; no additional programming or beacon signals required.
Initially, the control electronics will be Arduino; we plan on switching to a beagle board once we get the hardware tested and the basics out of the way. Robot porn to follow:
left to right top to bottom: gripper and wrist rotate assembly, rear view of motor heads, top view, side view - arm extended, close up of wrist rotate, mounted control electronics (Day 2)
RSS Feed