Goal 1: Create an autonomous, indoor roverbot to be used as a platform for other experiments in machine learning, human interaction, data collection, etc. Sub-goals include:
- Safely navigate an indoor environment. (Not necessarily perform SLAM)
- Self-charge. (I'm not starting with an iRobot Create because I'd like a more scalable platform. Plus, self-charge is a challenge.)
Goal 2: Add capability to perform data collection and store feedback from its environment by remote control over the web, including things like:
- Trigger by motion or noise
Goal 3: Add capability to manipulate objects in the environment via arm or gripper.
- Determine the weight of an object in its grip.
- Attempt to pick up an object and take pictures of it from several angles.
Goal 4: This is the pie-in-the-sky goal. I'd like to bring all the functionality together and allow the robot to rove around its environment and collect data about "objects of interest". It would compare the attributes of an unknown object against stored attributes of known objects in its database, and store the new object as similar if so. Generally, having the robot say "Hey what's that thing?" and then carrying out its own process for identifying it. As I understand it, this is the "correspondence problem" in robotics and has not yet been elegantly solved.
I've got a few more goals, but these should keep me busy for the rest of my life so I'll cover them later. Using the MS Kinect to perform SLAM would be awesome. I've found a few great resources for parts and info, but if you guys have any tips or general guidance for my project it would be much appreciated. There is still much to learn about the Force, and I am but a young padawan.