Realtime Robotics helps solve motion planning problems by providing robots with collision avoidance capabilities. Here, the robot avoids a researcher's waving hand.

George Konidaris still remembers his frustrating introduction to robotics: "When you are a young student and want to program a robot, the first thing that shocks you is how incredibly disappointed you are in your ability to use the robot."

Most new robotics experts hope to program their robots to solve interesting, complex tasks—but it turns out that simply moving them in space without colliding with objects is much harder than it sounds.

Fortunately, Konidaris hopes that future robotics experts will have a more exciting start in this field. That is because about four years ago, he co-founded Realtime Robotics, a startup dedicated to solving the robot "motion planning problem."

The company invented a solution that enables robots to quickly adjust their paths to avoid objects while moving toward a target. The real-time controller is a box that can be connected to various robots and deployed in dynamic environments.

"Our box simply runs the robot according to the customer's program," explains Konidaris, who currently serves as Chief Robotics Officer at Realtime. "It handles movement, robot speed, obstacle detection, and collision detection. [Our customers] just need to say, 'I want this robot to move here.'"

The key startup technology of Realtime is a unique circuit design that, when combined with proprietary software, provides robots with a plug-and-play motor cortex. In addition to helping meet the high expectations of robotics experts, this technology represents a fundamental step forward toward robots that can operate effectively in changing environments.

Realtime Robotics helps robots get around

Konidaris was not the first person to be discouraged by the motion planning problem in robotics. Researchers in this field have been studying it for 40 years. During his four-year postdoctoral fellowship at MIT, Konidaris worked alongside Thomas Lozano-Perez, a pioneer in the field who published papers on motion planning before Konidaris was born, in recognition of excellence in teaching engineering.

Humans take collision avoidance for granted. Konidaris points out that the simple act of taking a beer from the refrigerator actually requires executing a series of tasks, such as opening the refrigerator, reaching your body inside, avoiding other objects in the refrigerator, and deciding where to grab the beer can.

"In fact, you need to compute multiple plans," says Konidaris. "You may need to compute hundreds of plans to take the required action. ... Strangely enough, the simplest things humans do hundreds of times a day actually require a lot of computation."

In robotics, the motion planning problem revolves around the computational power required for frequent testing as the robot moves in space. At each stage of path planning, these tests help determine whether various tiny movements will cause the robot to collide with surrounding objects. Such tests have inspired researchers to think about increasingly complex algorithms in recent years, but Konidaris believes this is the wrong approach.

"People try to make algorithms smarter and more complex, but this often indicates that you are going down the wrong path," says Konidaris. "In fact, it is not common for advanced technology using advanced techniques to solve problems like this."

Konidaris left MIT to join Duke University in 2014, but he continued to collaborate with researchers at the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL). Konidaris is also where Duke met Realtime co-founders Sean Murray, Dan Sorin, and Will Floyd-Jones. In 2015, the co-founders collaborated to create a new type of computer chip with circuits specifically designed to perform frequent collision tests to enable robots to traverse space safely. The custom circuits can perform operations in parallel to test short-motion collisions more efficiently.

"When I left MIT for Duke University, what bothered me was that this motion planning problem should really be solved now," says Konidaris. "This really comes from a lot of experience at MIT. Before entering MIT, I would not have been able to write a paper on motion planning."

The researchers founded Realtime in 2016 and quickly hired robotics industry veteran Peter Howard, who is currently the CEO of Realtime and is also considered a co-founder.

"I wanted to establish this company in Boston because I knew MIT and a lot of robotics work was being done there," says Konidaris, who moved to Brown University in 2016. "Boston is a hub for robotics. There is a large amount of local talent here, and I think this is largely because MIT is here—PhD graduates from MIT became teachers at local schools, and then these people started robotics programs. This network effect is very strong."

Removing robot constraints

Today, most of Realtime's customers are in the automotive, manufacturing, and logistics industries. Robots using Realtime solutions can perform all kinds of tasks, from spot welding to inspection and picking items from bins.

After customers purchase the control box from Realtime, they load a file describing the robot's work cell configuration, information about the robot (such as its end-of-arm tooling), and the task the robot is performing. Realtime also helps optimally place the robot and its accompanying sensors around the work area. Konidaris says that Realtime can reduce the robot deployment process from an average of 15 weeks to one week.

Once the robot is up and running, the Realtime box controls its movement, providing it with instant collision avoidance capabilities.

"You can use it for any robot," says Konidaris. "You tell it where it needs to go, and we will handle the rest."

Realtime is part of the Massachusetts Institute of Technology Industrial Liaison Program (ILP), which helps companies connect with larger industrial partners. Recently, it joined ILP's STEX25 startup accelerator.

With plans for several large-scale deployments in the coming months, the Realtime team is driven by the belief that solving fundamental problems such as motion planning will unlock many new applications in the field of robotics.

Konidaris said, "What excites me most about Realtime is that we are a genuine technology company." "The vast majority of startups aim to find new applications for existing technologies; often, there is no new application or website, and not even a new 'vertical' where robots truly push technological boundaries. However, we have indeed invented something new, and this advantage and energy are what drive us. All of this is very MIT to me."