They did not sell enough units… but the real question is why they did not sell.
I think there were two primary issues with Rethinks offerings.
The main one was that the hardware was just not right for industrial applications. The compliant by design joints and relatively low torques and speeds just did not measure up to the competition such as ABB, Fanuc, Kuka or Motoman. Even the new generation of collaborative robots from Universal Robots, Aubo and Techman are faster, more powerful, more accurate and look and feel more industrial. The prominent buttons on the arm and plastic coverings looked fragile and like they would get bashed by whatever cables and tooling were nearby.
Engineers who had worked with both the Baxter and Sawyer told me in no uncertain terms that they did not think they were up to scratch in terms of hardware for industrial applications. They were unhappy using them again or recommending them to customers. Although it should be said the collective noun for robot engineers is a ‘whinge’.
This all feeds into the second issue. Industrial robot customers want return on investment, simple as that. So you have two competing requests, cheap or reliable. (Rarely but occasionally both –)
Reliability wise, as mentioned, many robot engineers had their doubts, but customers too could see little benefit in an industrial setting. Why not buy an ABB, Fanuc, Kuka, Universal or whatever that had proven itself with over 1000 times as many installs? I don’t know what the quoted MTBF (mean time between failure) was for rethink, or if they quoted it, but an ABB should run for 50,000 even 100,000 hours with nothing more than an oil change every few years. As I understand it Universals have gearboxes that have >30,000 hour MTBF that’s about 10 years of daily 8 hour shifts, and it is unlikely that the joints will be moving 100%.
Even if rethink had comparable figures for reliability you have to have an incentive for customers to take that risk – and they are very risk averse. This means you have to compete on costs or features. Newcomers to the market such at theand are less than half the cost of their competition to overcome this hurdle.
The Baxter and Sawyer were not crazy expensive but they were very comparable to the competition with headline costs much the same as any ABB or universal. In terms of overall install costs they would likely run out more expensive than most as without a significant install base whole solutions would have to be developed from scratch. Compare this to a Fanuc or ABB where a robot engineer can reference hundreds of installs over the last 30 years for almost any application. Integrators would need to add in development time and budget when using a Rethink. So cost then was a real problem.
The added features such as being safe to work with, emotive screen and hackable interface were great for Universities but are of little interest to industrial users. The safety thing is a point I have made on many occasions. Ok the robot won’t damage you significantly when it hits you, but is that really ok? If it just bruises you a bit every few minutes is that ok? No it is not. You don’t want to be repeatedly bashed by a robot so it needs to stop before it does that. Being a bit soft doesn’t help much. The screen is fine, but so is a cheap HMI that can be put exactly where it is needed to communicate. Hack-ability? In industry? No thanks.
Comparatively unsuited hardware for industrial market.
Cost to risk too great for industrial customers compared to competition.