An Instructables post by user PenfoldPant, a laser-guided "ghost climber," caught my attention today as a really cool project worth discussing.
Penfold explains that as an amateur rock climber, he frequently sees other climbers climb a route and wonders if he can do that route, too. However, he can't always remember the route the other person took when he tries to duplicate it. His solution is a laser pointer on a turret. Two stages are used: a recording stage, in which an operator keeps the laser pointer aimed at the climber as the climber climbs, and a playback stage, where the laser pointer replays the recording so a second climber can attempt to duplicate the route.
While I think the project is uber cool, it got me to thinking how I might solve this problem.
I see two main issues with the project. The first issue is that during the record phase, only one point on the climber's body is tracked. I'm not a climber, but I imagine that if I were trying to duplicate another climber's route I'd want to know which holds he used for each of his four limbs. (Can anyone correct me if I'm wrong?)
The second issue is a little bigger. During playback of the route, the laser pointer is projected in the same place it was during the record phase, which is on the climber's body, which means it can't be seen by the climber trying to duplicate the route!
That, my friends, is an engineering challenge! So.. how would I approach the problem? Well, I'd want to track 4 points on the climber during recording: each wrist/hand and each ankle/foot. Maybe use an IR LED fastened by velcro? A colored ball? I think the solution would involve some kind of image/video recognition which admittedly is not easy. As luck would have it, Zdenek Kalal recently released his "Predator" video object tracking software as open source software; that could certainly be part of a solution.
Once the route is recorded, motion analysis could find the holds the climber used by looking for places where the limb stops moving for, say, a second or so.
During playback, a second climber could wear the 'limb trackers' and colored lights projected onto the wall could be used to indicate which limb goes where. The computer could even automatically determine when each limb is on the correct hold (because the second climber is also wearing the limb trackers) and automatically advance the indicator to the next hold for that limb for practice, or play back the recorded route at recording speed for competitions.
This system would solve the playback issue as well, since the projected target would be where the climber's limb should go and would probably not be occluded by the climber's body in most cases. (Of course, there will always be instances where the climber's body occludes the sensor/projector. Addressing that issue is an interesting thought experiment in itself.)
While it sounds like a fun project, it's probably not one that I'll ever have the time or inclination to address. It is a fun thought experiment, though.
How would you attack this project?