The challenge specified that the robot must be off the ground and must lift itself only using the handle (2.5"x0.25"x0.5"). Limited by 8 total motors, my team constrained the design to use only 2 of the 8. The lock mechanism, a deadbolt-like mechanism, was designed first, followed by the rack-and-pinion mechanism. It's likely that this chronology led to the design's reliance on rack-and-pinion, which was needlessly complex compared to other teams.
Most notably, the flimsy 3d-printed racks were replaced with 1/4" steel racks via a CNC plasma cutter. Additionally, the second stage of gearing was added so that the effective gear ratio of the lock riser was 4:1 (4 times as much torque). Both these changes required the complete redesign of the rack housings. Within this new design, the upper housings' gear container was modified so the exposed face was oriented towards the motor. This exposed face allows easy access to the gearing. A cover, which holds the motor is bolted on to fully contain the gears.
Macro-View of Lock Riser V4 and Lock Mechanism Illustrates Internal Gearing and Pinion
Through each iteration of the design, the size of the mechanism grew. The size and complexity of the mechanism were a major nuisance. The extrusion on the gear container cover where the motor mount attaches was still too weak; it worked, but with use, the plastic began to crack, requiring the eventual replacement.
Both issues (not strong enough and the weak motor-mount support) were addressed with the redesign of the upper rack housing. The new design geared down the motor to increase the torque, by using worm gears. I'd never used worm gears, but they are known for being compact in terms of space used versus the gear ratio. The new gear type moved the motor parallel to the rack housing, meaning the force acting on the housing during lifting/lowering could be better distributed.
The worm-gear system proved to be a major flop; the 3d-printed worm gear broke at its point of attachment (i.e. between the base and the point of attachment). Perhaps gears made of metal or a strong plastic such as nylon would have worked.
Both issues (not strong enough and unwanted rack housing movement) were addressed with the redesign of the upper rack housing. The new design geared down the motor to increase the torque; it also modified the slots in which the racks were slid into.
The lift still wasn't strong enough. (In retrospection, I should've taught myself the physics acting in the system and base the design off calculations, as I did here). The other issue was the extrusion on the upper rack housing that supported the motor was too weak for the forces applied to it during lifting/lowering.
The lock riser mechanism is used to lift/lower the lock mechanism during attachment and raise the robot of the ground during the lift phase. The rack-and-pinion was used for two reasons: One, my relative familiarity with rack-and-pinion. Two, the existing lock mechanism was designed with the assumption that it would be vertically placed on the lander's handle.
The first version of the mechanism is a "direct drive" system in which herringbone teethed racks and pinions were directly attached to the motors and lock mechanism respectively. The housings for the racks were designed in two pieces to ease 3D-printing (print volume). The base of the housing was attached to the grid plates, which spanned our robot.
First, the two motors weren't strong enough to lift the robot. Second, the cutaway material on the upper rack housings (allow lock riser to collapse) was too loose allowing the rack to move in directions other than strictly vertical.
The primary design consideration taken into account was the robot not moving once locked. In retrospection, taking into consideration other teams' incredibly simple designs this was an err. The deadbolt lock mechanism proved effective when it was carefully aligned; however, carefully aligning the robot via remote control is easier said than done.
The body was designed in two pieces, which screw together via threaded-brass inserts. The 2 part design allows the deadbolt to slide within the grooves of the first body piece while being limited by the second. The desired operation is as follows: The driver of the robot raises the lock mechanism. They align the funnel below the deadbolt with the handle on the lander. The drive forward and lower the mechanism. Finally, the deadbolt is closed attaching the robot to the lander, which can then lift itself up. The deadbolt is slid open and closed via a hobby servo. The rotating motion of the servo translates into linear motion of the deadbolt via the connecting-arm which pivots on each end where it is attached.
The mechanism is assembled of many smaller parts, which makes replacing parts (because of design update(s) or break(s)) easier and less time-consuming. The design looks cool. However, as the designer, the issues far overshadow anything it does well. First, one of the integral issues is that lining up the mechanism with the handle proved incredibly difficult. Second, the servo means that a set of wires are hanging off the back of the mechanism--increasing the number of things that can go wrong (not to mention the extra slack needed for when the lock is raised). Both of the previous issues could be solved with a simplified, purely mechanical design. Finally, the deadbolt, which slides in an extruded dove-tailed grove (right side of photo above--one can see the lower groove below the white connecting-arm), had an issue where the front of it (left in photos) would "pop" out of the mechanism's body; this was because only the back of the deadbolt was anchored. To fix this, I would implement a catching feature, meaning the front of the deadbolt is funneled into a groove in the main body piece as it spans the lander-handle-gap.