Sramana Mitra: Your point is well taken. From what I know from bringing products to market, anything that requires too much work on the part of the consumer basically fails. As long as the products that come onto the market have self-learning capabilities, that would be fine.
Rich Mahoney: At the end of the day, these robots will be products. They will have to meet customer demands and have some value for the price people are paying for them. That is a phase that has to happen. There is a lot of attention to robotics right now. Even the smallest bits get a lot of attention at the moment. That is just the nature of the technology. But if you compare it with any other consumer product or any other area, the overall activity is extremely low. It is still very early.
SM: The comments you made earlier about medical robotics are unambiguously promising as well as high value. Complex surgeries may be done through a control system by surgeons, but the actual manipulation is done by robots. That is a great use case. I don’t think there is any ambiguity on how that should scale, as long as the price points are reasonable.
RM: I have been involved in the medical device field for a while. For many medical devices, when they enter the market, there are challenges in terms of adoption, training, and making sure that a device is fully understood. Robotics is going through a bit of that phase right now. There are more court cases around da Vinci, there is more training and other things. That is the normal course for that kind of new technology.
I don’t think robotics will go away because of this. I think the price is a big barrier right now, and eventually surgical and other medical robots will benefit from the work that is going on to create higher performing robots at lower price points. I introduced a robot for stroke therapy in my previous job. That was a robot that does a similar activity as a therapist does when working with someone to help them recover their arm function following a stroke. The challenge there is that in the current clinic goal setting, a therapist gets an hour a day with the patient to work on their arm function. But the patient also need to work on walking, speech, and other activities of daily living. There is not a lot of time to focus on the arms.
There is the idea that therapists are very good at what they do, but the robot can just focus on one thing. However, the robot can also collect information not just about how the patient is moving, but the also about what it was programmed by the therapist. So the therapist is saying, “I want you to do 100 [repetitions] of that type of movement and then this type of therapy.”
You are also capturing the critical decision making of the therapist at the same time. You had this machine that was interacting with people, and it was capturing information on the patient, on the therapist, and ultimately on the clinical outcome of the patient. Now you have this way of understanding the quality of the therapy relative to the outcome you are getting. If you start to think about networking those robots and capturing that information broadly, you have a way to set norms, where a therapist could be saying, “I have this particular patient in front of me. Let’s see what the norms look like for this patient. What is the recommended training approach?” That approach would be based on potentially tens of thousands of therapy sessions, thousands of patients, and thousands of clinicians making their best decisions. Over time you could imagine that the actual decision making would be narrowly defined in terms of the best outcome for a patient. Robotics will work that way in many environments, where we begin to understand the quality of the service, capture it over time, and then use the robot to deliver that quality service in the same platform.