I attended the BYU CS Department Colloquium today. The speaker was David Robinson from Seqway LLC. David is a BYU grad who got a PhD from MIT and then went to work on the Ginger project which eventually lad to the development of the Segway Human Transporter. David's in the Core Technology Group. The talk is about the technology behind the Segway HT.
Balance is the easy part. The Segway is an inverted pendulum which is a classic problem appearing in Chapter 1 of most control textbooks. The motors not only provide the motive force, but also the torque necessary to balance the machine. The problem is that there is a speed where the motor has no torque left and looses it balance. The trick is to figure out how to limit the torque-speed curve so that users don't fall down.
The battery provides sufficient power for the Segway to climb fairly steep ramps. more interesting, the Segway regenerates power to the batteries when the machine goes downhill. This is good on two fronts. First, you get longer battery life. Second, generating power causes a braking effect, when needed.
An interesting aspect of the design was creating the gear train in such a way that it sounded quite. Part of making the acoustics right was picking gear ratios to that the noise that the machine makes is "in tune." That is, the two primary sounds that the gear train makes are two octaves apart. David says that getting it wrong really sounds bad.
The processor is a TI2406 with 32K of memory and fixed point math.
The battery last about 12 miles depending on terrain. There have been some issues with batteries causing system failure (i.e. people fell down). The recent recall fixes the problem. David reemphasizes that this is a machine, not a magic carpet and it does have limitations.
David spent a fair amount of time talking about Segway's cultural principles. I thought they were good advice for an engineering team.
- Go fast. Mario Andretti said "If you feel like you're in control, you're not going fast enough."
- Expect conflict. The only way to truly tolerate design failure is to do it fast and early.
- Expect to fail. Be tough on issues, easy on people. Cross-functional conflict must happen early to work out the right ideas.
- Have fun.
- Let people be different. Ideation people vs. execution people. (Innovation vs. delivery)
- Hire the best. Jeff Bezo's said "A's hire A's; B's hire C's; C's hire D's"
- Be humble as an individual; be proud as a team.
Innovation is hard. You spend a lot of time out of your comfort zone.