It's a cool idea. It might be a little too formal for the kids. I mentor on a FIRST Robotics Challenge (high school version) and it's tough even getting the high schoolers on board with process.
I like the idea of the missions being the backlog. One of the objectives/judging criteria is how the teams strategized on what missions to choose. And prioritizing the backlog would be a great way to talk to that.
For Planning Poker, have any of the kids done this before? Estimating is tough even for grown ups :). I like the idea of showing that tasks take longer than expected though. Half an hour seems fine. You might even do 15 minutes, 30 minutes, 1 hour to allow for some small tasks. Like cleanup :). Or multiply the # kids on ask by the time.
Who decided which missions you choose before Scrum? I like the idea of the kids deciding what to do. You can be a trusted advisor rather than the product owner. I think it would be better for you to be ScrumMaster than product owner as that is more of a coaching role.
Of course it is fine for students to be testers and developers! On a real Scrum team, the goal is for people to pitch in and do whatever is needed. That's the cross-functional team objective.
I think a sprint length of 4 weeks is good. One week for sprint planning/retrospective and three for doing actual work. I hope you'll post how this went. (While that's more discussion than StackOverflow, I highly recommend you post your experiences on Chief Delphi (a FIRST robotics forum). And if you do, please add a comment with the link!
You could do test first design. What is the definition of success for a mission? For example, last year the robot had to drive straight, turn left and toss a ball at bowling pins. A test could be "make sure the bowling pins are knocked down." This real world intro to test driven is cool!
My general advice would be to try out parts of Scrum. Some are likely to be a better fit for kids than others.