Investing in Dexterity

Freeing robots from cages, and pioneering the next generation of industrial robotics

Nan Li
Obvious Ventures
4 min readJul 21, 2020

--

Industrial Robotics is currently a $40B global market, growing at double-digit rates per year as a wide variety of industries drive initiatives around digital transformation. Despite a rich ecosystem of OEMs developing significant improvements to hardware over the past few decades, the core cognitive capabilities of robots have mostly remained the same during that time.

The vast majority of robots in the field today are meticulously programmed to follow a static set of instructions to complete some specific task.

These robots are set up with custom code that dictate point-by-point guidance for that task, and then left to repeatedly perform that exact set of instructions until the task changes and the robots need to be reprogrammed again. This open-loop controls paradigm creates robots that are extremely limited in a number of ways:

  1. Long setup times: Custom development for each project can take upwards of nine months.
  2. Frequent shutdowns/servicing: Robots follow hard-coded instructions and break when changes are introduced to the task or environment.
  3. Limited human safety: Humans need to be separated from robots due to lack of contextual awareness.

These critical limitations keep robots locked away in cages, constrained to jobs that require no judgement, and that stay unchanged for long periods of time.

Caged Robots at Tesla’s Fremont Factory (Source: NYT)

Unsurprisingly, the vast majority of manufacturing and supply chain jobs do not meet that specification. In these facilities, most tasks require real-time decision-making based on perception of the environment and awareness of other workers. Most tasks also go through constant change as products being manufactured and shipped evolve.

To meet these conditions, the robots of tomorrow will need to have heightened perception, real-time cognition, contextual awareness, and human safety. They will need to make decisions about how to best perform the job at hand and be able to make adjustments as new scenarios come up.

These capabilities will allow next-generation robots to perform a broad set of tasks that will unlock an order of magnitude expansion of the industrial robotics market today.

This future cannot come soon enough. Rising global consumption, additional complexity caused by eCommerce growth, and most recently, an increased emphasis on resiliency in light of COVID-19 have all put enormous strain on the global supply chain. Facilities frequently struggle to hire and retain employees due to the tough working conditions and repetitive, menial work. We believe that robots will continue to automate dull, dirty, and dangerous work to enable humans to have higher-order, more strategic roles in these facilities.

Enter Dexterity

I was referred to Dexterity’s CEO, Samir Menon, in 2017 because he was a thought-leader in robotics research through his PhD work at Stanford. The promise of next-generation robots with human-like intelligence and agility had already propagated through the industry for many years, with mostly disappointing results. I had the opportunity to meet with many robotic startups during that time, but had yet to find a company that was developing compelling technology and showed a strong sense of empathy for the problems facing the industry today.

In a short period of time, Samir demonstrated both of these important qualities in spades. First and foremost, he had a strong point of view on various technological approaches to robotics. However, his counterpoints to many of the approaches were rooted in pragmatism — how they wouldn’t work in the real world. Even though Samir was still wrapping up his PhD work at the time, he was already deep in forming the foundational tenants of what would eventually become Dexterity:

  • Dexterity takes a full-stack, holistic view to robotics instead of overly relying on a specific technology like perception or reinforcement learning. The company has assembled a team with decades of expertise in areas of robotics ranging from force feedback, physics modeling, deep learning, motor controls, and human-robot-interaction. They believe that the best solution draws from many of these fields simultaneously to create precision and redundancy.
  • Dexterity’s capabilities are hardware-agnostic and highly modular. The company has already deployed software across a wide range of applications that call for different types of robots, end effectors, and work configurations.
  • Dexterity delivers solutions, not technology. The company packages all of the capabilities together to bring customers Robots-as-a-Service. The RaaS model allows customers to capture immediate ROI on projects and aligns incentives for both sides to scale up deployments vs. spend time in demos and pilots.

Obvious jumped at the opportunity to invest back in 2017 and it has been amazing to watch the team progress over a relatively short amount of time. Through a lengthy 2.5 year stealth period, Dexterity has quietly built out one of the best teams in industrial robotics, deployed live projects alongside some of the world’s best supply chain operators, and raised over $55M between a Seed in 2017 and a Series A last fall.

We believe that the next decade will be a golden era for industrial robotics as technological advancements finally deliver on the decade-long call to action. Dexterity is leading the charge as a new wave of companies make a lasting imprint on the industry.

Excited to officially welcome Dexterity to the Obvious portfolio and support them on an already exciting journey!

--

--

Nan Li
Obvious Ventures

GP @ Obvious; technology, music, culture enthusiast