Northsence

Whenever I say: “Most of the simple tasks are already automated. What is left is complex tasks in unstructured environments,” I get a lot of backlash. But let me explain the same idea from a robotics systems perspective: system complexity vs robotics versatility. This is purely my own perspective, but I think it is a useful lens for practitioners, investors, and anyone trying to understand robotics without starting from hype, AI, or humanoid headlines. Classical robotics worked by narrowing the problem. Industrial arms, AMRs, and simple grippers solved specific tasks in structured environments. The robot did not need to understand the messy real world because the world was redesigned around the robot. Fixtures. Conveyors. Safety cages. Known parts. Known paths. Repeatable workflows. The complexity did not disappear. It was pushed into the environment. Modern robotics is moving in the opposite direction. As robots are expected to work in less structured environments, the machine has to carry more capability internally: mobility, manipulation, perception, dexterity, autonomy, recovery, and safety. This is why robotics is moving from specialized systems to converged systems. A mobile base gets an arm. A gripper becomes adaptive. A cobot becomes bimanual. A quadruped gets wheels. A humanoid combines mobility, hands, perception, and task planning. Etc The goal is versatility. But versatility is not free. Every added capability increases integration complexity, failure modes, maintenance, cost, and validation effort. So the real question is not: “Is this robot advanced?” The better question is: Does the added complexity create useful, reliable, and economically meaningful versatility? That is also why I think the “humanoid vs industrial robot” debate is often framed too simplistically. (Also “my bad” here) It is not about one replacing the other. A fixed industrial arm is still the right answer when the task is stable and the environment can be structured. A more “converged” robot starts to make sense when the task is variable, the environment is harder to control, or the cost of redesigning the world around the robot becomes too high. So for me, the real question is: Where should the complexity live? In the environment, through fixtures, process design, and dedicated automation? Or inside the robot, through mobility, manipulation, perception, dexterity, and autonomy? Because complexity alone is just burden. Versatility is the prize. (as I tried to explain the image, but take it with a grain of salt) This is only the hardware system, the same logic can be applied to Software, foundational model and Human-robot Interaction. If you want to read more, check out my latest blog. #humanoids #robotics #robotdesign #physicalAi #cobots #AMR #VLA