Case Study: Rapid Robotics
Last updated
Last updated
→ www.rapidrobotics.com Archetypes:
Wave 1: Structured environments; Stodgy industries left behind
Wave 2: Companies balancing off-the-shelf and custom built solutions, innovating in speed to implementation; Startups building optimized infrastructure to extract value from emerging and maturing data sources
Wave 3: Disintermediation of the hardware and software stack; Network effects; Leveraging an as-a-Service (“aaS”) business model; Enabling out-of-the-box automation
Opportunity Areas:
Wave 1: Mechanization of Prevalent, Procedural and Time-Consuming Tasks
Wave 2: Bundling Hardware and Software to Extract Value from New Data Sources; Mutable Software Wrapped Around Off-the-Shelf Hardware
Wave 3: Going Big with Automating High Frequency Tasks
Vision Rapid started by defining an “automatable” process, by identifying the rate-limiting step in the adoption of industrial robotic arms by the “Mighty Middle” manufacturing market: integration into the factory. The team paired a new computer vision dataset with existing, off-the-shelf hardware and raw materials. The team then transformed a service-oriented, expensive, and months-long system integration effort requiring third parties into a cost-efficient, quick and repeatable (read: proprietary) cycle.
North Star On the path to market, the north star was always the customer journey. Rapid prioritized solving for a problem rather than perfecting the technology to then find a problem to solve with it. They engaged in intensive customer discovery: listening, watching, and communicating constantly.
First Use Case Rapid's first use case was automating the task of machine tending. Importantly in that first deploy, they built a clear path to programmable digitization for flexible, no-code task expansion. But how? As a task is automated by one robot, it is saved as an application that can then be used by any robot across any production line. This unique task library data set allows Rapid to deploy dumb hardware and program it to act intelligently.
Plumbing needs to be in place to leverage a new dataset’s “gold mine”, and get the data flywheel spinning. Rapid combined off-the-shelf robotic arms, custom-made grippers, and a custom computer vision system with their task software layer.
To achieve the desired performance of the first use case, Rapid needed to build interdependent components of the Rapid Machine Operator (RMO): software, hardware, networking, and customer service.
1) Data & ML Infrastructure to Analyze the New Data Set, and Perform Compute
Rapid’s cameras collect terabytes of data, which need to be processed and analyzed by deep learning models that allow for the real-time output control of the arm and “machine vision”.
To productionize algorithms, Rapid converted generalized perception algorithms into functional products built for industrial environments.
To ensure adoption, Rapid provided a critical user-friendly interface and experience.
2) Hardware and Actuation Rapid mixed off-the-shelf, customized, and custom-built solutions:
Custom-built: the self-contained “work cell” or RMO
Customized: gripper, and computer vision system
Off-the-shelf: robotic arm, air regulator and iPad
3) Networking Rapid's cloud-based task library allows for a centralized system of intelligence across customers, and provides Rapid the capability to manage individual RMO units, and push software updates remotely.
4) Customer Servicing Rapid guaranteed customer servicing (troubleshooting & debugging) for any and all machine issues. This step will become increasingly automated, and perhaps one day monetized independently.
Rapid’s true moat lies in moving from a task-management solution to a task-management platform. This wave is divided into two steps:
Nailing the First Use Case with a Product in Market This outcome is the culmination of steps taken during W1 and W2. At the get-go, things never go exactly as planned and that’s exactly the reason to rapidly iterate. Rapid experimented tirelessly until the RMO was production-ready and built in the name of customer satisfaction.
Once the First Use Case Earns Product-Market Fit, Expansion becomes Horizontal For Rapid, this meant moving beyond machine tending to tasks like pad printing, heat stamping, and more. As the software gets “smarter” through iterations of W1 and W2, more complex use cases can be explored beyond the low hanging fruit. Every task that’s built is saved in a centralized library, where it can automatically be re-used by any robot across any production line.
The data flywheel doesn’t stop spinning: more tasks lead to wider adoption that leads to a broader, data-driven task library that leads to more operating data that leads to more tasks. With these persistent improvements, scalable financial gains are unbounded.
By adopting a customer-centric approach, circumventing the Systems Integrator, and iterating at speed through the 3 Waves, Rapid is realizing its vision to become the Salesforce of the manufacturing industry and bring back the possibility of “Made in America”.
