Scaling safe and effective procedures
Surgeries by humans are limited by our motor skills and number of doctors.
Autonomous science is an inevitability and so is autonomy in other settings like surgery. Tracking of general robotics progress shows a breakneck pace with RT-2 vision-language-action, Octo which performs similarly to RT-2, pi-0 trained on the largest robot interaction dataset, GeRM which mixture of experts and off-line reinforcement learning, and even diffusion models for bimanual manipulation of robots. We’ve tracked companies like Matic which bring intelligence to vacuums (and consumers).
While things such as the Da Vinci Robot have been in hospitals for years, the number of new entrants is relatively low. In some ways, this makes sense because hospitals are hard to sell to, doctors hard to replace, and too overloaded to change the status quo. However, there have been some fairly large breakthroughs and efforts we should be aware of and keep tracking. There are surgeries that surgeons really struggle with but could be valuable if we employ medical robotics. Companies building likely need to be general purpose and generally trained, while initially focusing on intractable or difficult surgical procedures. Some examples here:
Other Thoughts
