What 10+ Years in Robotics Teaches You

Robotics careers

Seven hard-won lessons from people who have spent decades building, breaking and deploying robots.

By Chris Escobar · 9 minute read

Robotics has no shortage of impressive demonstrations.

A humanoid completes a new maneuver. A mobile robot navigates a difficult environment. A robotic arm performs a task that previously appeared impossible.

But experienced roboticists know that a successful demonstration is only the beginning.

Turning that demonstration into a dependable product introduces an entirely different collection of problems: safety, manufacturing, reliability, integration, maintenance, customer expectations and the unpredictable behavior of the real world.

Drawing on public interviews and essays from people who have spent more than a decade in robotics, here are seven lessons for the next generation.

If you aren’t breaking robots, you may not be learning enough

Boston Dynamics founder Marc Raibert has spent more than 40 years working on legged robots. Much of that time has involved deliberately finding the limits of what those machines can do.

“Build it, break it, fix it.”— Marc Raibert, founder of Boston Dynamics

Inside Boston Dynamics, a new capability is pushed until it fails. A robot may fall dozens of times before producing the successful movement eventually seen by the public.

The lesson is not to celebrate careless engineering. It is that failure, when deliberately produced and carefully studied, is information.

A robot that never encounters its limits inside the workshop is likely to discover them for the first time in front of a customer.

For a young roboticist, learning to diagnose failure may ultimately be more valuable than learning to produce a flawless demonstration.

Read the original interview on CBS News →

Lesson two

Deployment is a discipline of its own

Melonee Wise has worked in robotics since joining Willow Garage in 2007. She later co-founded Fetch Robotics and helped move autonomous mobile robots into commercial warehouse environments.

“It takes 10 years to have an overnight success.”— Melonee Wise, robotics entrepreneur and former Fetch Robotics CEO

Wise argues that veteran roboticists become pragmatic because they have learned how different building a robot is from deploying one.

The most difficult work may not be perception, navigation or manipulation. It can be fleet management, IT security, compliance, customer integration and the operational systems surrounding the machine.

A robot does not enter an empty environment. It enters an existing organization with employees, workflows, safety requirements, software and expectations.

The younger engineer naturally concentrates on making the robot perform its task. The veteran also asks what happens when 50 robots are performing that task, one loses connectivity, another needs maintenance and a customer changes the surrounding workflow.

That broader operational view is what turns robotics technology into a robotics business.

Listen to the Melonee Wise interview →

Lesson three

Go into the environment before deciding what to build

Andrea Thomaz began working in social robotics during the early 2000s. When she and her team at Diligent Robotics started developing Moxi, a hospital robot, they resisted assuming that robotics expertise made them experts in healthcare.

“We knew that we didn’t know enough about nursing.”— Andrea Thomaz, co-founder of Diligent Robotics

Diligent spent months placing an early prototype inside nursing units. Engineers and product staff worked alongside hospital personnel, asking what they actually needed a robot to do.

This fieldwork changed the product.

The original robot had open containers. Once it entered hospitals, the team discovered that customers needed to know who had removed an item. The eventual system incorporated locking storage, authorization and an employee badge reader.

This was not a problem that could have been uncovered entirely in a robotics laboratory. It emerged only when the machine met its working environment.

The lesson for the next generation is to treat users as participants in the engineering process—not simply as people who receive the finished product.

Read the Andrea Thomaz interview on Robohub →

Lesson four

Begin with the human problem, not the robot

Gill Pratt’s career includes MIT’s Leg Lab, the DARPA Robotics Challenge and leadership of the Toyota Research Institute.

His advice is to start with the person who will use the technology.

“Start with the human being, the user.”— Gill Pratt, CEO of the Toyota Research Institute

Pratt uses elder care as an example.

A young engineer might imagine a robot that prepares an older person’s meals. But cooking may provide that person with independence, purpose and a way to contribute to family life. The better robot might help the person continue cooking rather than take the activity away completely.

Both machines could be technically sophisticated. Only one addresses the deeper human need.

This distinction matters far beyond healthcare. The most advanced robot is not necessarily the most valuable robot. Value depends on whether the machine improves the life or work of the person using it.

The technical question is: “What can this robot do?”

The more important question may be: “What should this robot help a person continue to do?”

Read the Gill Pratt interview in IEEE Spectrum →

Lesson five

A prototype is not a supply chain

Rodney Brooks has spent more than four decades in robotics. He co-founded iRobot and helped take the Roomba from an idea to a mass-market product.

Brooks recalls that iRobot’s first Roomba manufacturing batch consisted of 70,000 units—already an extraordinary number for a robotics product. The company eventually reached millions of units annually.

That experience produced a practical lesson: robotics companies should use standard components and established high-volume supply chains wherever possible.

“Find someone else’s big juicy supply chain and grab a hold of that.”— Rodney Brooks, roboticist and co-founder of iRobot

A custom motor may give a prototype precisely the desired performance. A standard motor produced for an established market may provide better pricing, availability, quality control and replacement options.

Engineering for production involves different trade-offs from engineering a research platform.

Elegance is not just achieving the ideal technical specification. It is creating something that can be manufactured, supported and repaired at a viable cost.

Read “Tips for Building and Deploying Robots” →

Lesson six

Robotics rewards people who look outside robotics

When MIT roboticist Cynthia Breazeal developed Kismet, one of the pioneering social robots, she did not limit the project to conventional engineering expertise.

The work drew on developmental psychology to understand how people learn and communicate. Breazeal also consulted cartoon animators.

“How do you make something that’s not alive appear lifelike?”— Cynthia Breazeal, MIT professor and social robotics pioneer

Animators understood how small changes in timing, expression and movement could communicate attention or emotion. That knowledge helped inform the design of a robot capable of giving people understandable social feedback.

Modern robotics draws from mechanical engineering, electrical engineering, computer science, psychology, design, biology, safety and ethics. No individual can master all of them.

The skill is developing genuine depth in one area while remaining curious enough to collaborate across the others.

A roboticist does not become interdisciplinary by knowing a small amount about everything. They become interdisciplinary by knowing what they do not know—and recognizing whose expertise should be brought into the room.

Read the story of Cynthia Breazeal and Kismet at MIT News →

Lesson seven

Careers in robotics are rarely linear

Ayanna Howard joined NASA’s Jet Propulsion Laboratory in 1991 and went on to work on artificial intelligence, machine vision and autonomous systems for planetary exploration.

In a NASA profile, Howard summarized her approach simply:

“I just do what I love, and somehow the opportunities unfold.”— Ayanna Howard, roboticist, educator and former NASA researcher

This should not be mistaken for advice to leave a career entirely to chance. Howard built deep technical expertise and pursued work involving demanding real-world applications.

But robotics careers frequently develop through unexpected projects, collaborators and problems.

People enter the field through mechanical engineering, controls, software, AI, neuroscience, industrial design and many other routes. What unites them is often less a particular qualification than sustained curiosity about intelligent machines interacting with the physical world.

Young roboticists should certainly think about where the industry is going. They should also pay attention to the problems they cannot stop thinking about.

A field as difficult as robotics demands too much persistence to be chosen solely because it is fashionable.

Read the Ayanna Howard profile from NASA JPL →

What should the next generation take from this?

Taken together, these veterans offer a picture of robotics that is more demanding—and more interesting—than the one presented in polished demonstration videos.

Learn the fundamentals. Break things thoughtfully. Spend time with users. Develop expertise, but do not become trapped inside it. Understand manufacturing and deployment. Look beyond technical performance to the human result.

Above all, remember that the robot is only one part of the system.

The customer, workplace, infrastructure, support team and people living or working alongside it are part of the system too.

The next generation of roboticists will not succeed simply by building more capable machines.

They will succeed by building machines that belong in the world.

Robotics recruitment

Building a robotics team—or considering your next move?

Escobar & Associates works exclusively across robotics, autonomy and physical AI. We help ambitious companies find the people capable of taking difficult technology from research and prototype through deployment and scale.

If you are building a team, looking for specialist robotics talent or thinking carefully about the next stage of your career, we would be glad to hear from you.

Discuss your hiring plans • Explore robotics opportunities

Research note: This article is an independent synthesis of publicly available interviews, profiles and first-person essays. The individuals featured were not interviewed by Escobar & Associates for this article. Quotations have been kept brief and linked to their original sources.

In this article