Vinsa to build visual AI workflows with Boston Dynamics’ Spot quadruped robot

The Robot Report
Vinsa visual AI modeling

Vinsa’s Intelligence Models work with Spot to perform inspections, reading gauges and detecting areas of concern with high accuracy. Source: Vinsa

Boston Dynamics Inc., which is known for its legged robots, today announced that it has partnered with artificial intelligence and vision provider Vinsa. The companies will use Vinsa’s Intelligence Models to enable Boston Dynamics’ quadruped Spot and other robots for new use cases.

South Florida-based Vinsa (legally Recon AI LLC) was launched this year and designs and deploys custom visual AI systems. The company said its “Intelligence Models and Core AI Engine bring a human-in-the-loop, chassis-agnostic approach to implementing computer vision technology.” It serves industries including utilities, manufacturing, and construction.

Waltham, Mass.-based Boston Dynamics stated that Vinsa’s Intelligence Models can be deployed on Spot in 30 to 45 days. It said the combination of visual AI and its mobile robot can help companies reduce operational costs and realize incremental returns on investment (ROI) by automating labor-intensive tasks such as facilities inspection.

Michael Perry, vice president of business development at Boston Dynamics, and Daniel Bruce, co-founder and chief AI officer at Vinsa, responded to the following questions from The Robot Report:

Vinsa brings machine vision models to Spot

How did this partnership come about — how did your companies find each other?

Perry: Vinsa is one of more than 60 companies that have joined the Spot early adopter program. They were the first to identify semantic computer vision solutions running on the edge as a clear value-add for industrial customers.

Bruce: Vinsa was one of the first companies to deploy Spot in the field for a client.

How long have Boston Dynamics and Vinsa been working to integrate their offerings?

Perry: Vinsa started working on their solution several months ago, with their first commercial deployment of Spot using their software in January. Since then, we’ve supported their integration with Spot, ramped up deployment with additional pilot customers, and validated the use case of our integration.

What capabilities does Vinsa add to Spot that it didn’t already have?

Perry: Spot is built to navigate tough terrain, either autonomously or with a remote human operator, and it has been deployed to inspect a number of unstructured environments, ranging from construction sites to oil rigs and even nuclear reactors.

In many of these applications, one of the most dangerous or time-intensive tasks is inspection. With our Spot CAM payload, the robot can take detailed color photos on these industrial sites, but Vinsa is a key part of the workflow for transferring the data into something meaningful for the customer.

Vinsa can take those photos and make conclusions about the safety of a plant or find anomalies on a production line so that Spot is not only automating data-collection, but also triggering important actionable insights for operations managers where necessary.

Bruce: Vinsa adds AI to Spot’s vision-capturing capabilities, processing the visual data and turning it into critical information. Vinsa enables near-term ROI by taking visual data captured by Spot and turning it into action — automating laborious tasks, detecting anomalies or threats sooner, and optimizing inspection hours from humans. Vinsa helps the implementation of robotics and AI to pay for itself.

Vinsa to build visual AI workflows with Boston Dynamics' Spot quadruped robot

The Spot robot captures visual data in all kinds of facilities, helping to improve awareness of operations and to automate laborious tasks. Source: Boston Dynamics

Spot hardware and Vinsa software

How much of your collaboration is on the hardware side?

Bruce: Vinsa’s custom Intelligence Models run on Spot, which is the hardware component. These models include gauge reading, thermal imaging, leak detection, people tracking, PPE [personal protective equipment] tracking, and many other capabilities.

How much of the partnership focuses on software development? Is Vinsa’s visual intelligence connected in any way to Spot’s navigational systems?

Perry: Primarily, Vinsa’s computer vision models are being used to autonomously analyze the data collected by Spot.

Bruce: Spot is able to understand when it sees items like analog gauges; Spot can then walk up to the gauge, take a reading, and trigger the appropriate action based off that reading, all because of Vinsa’s Intelligence Models.

Is Vinsa working with other robotics or drone companies on mobile detection and analysis?

Bruce: Vinsa works with best-in-class robotics, drones, and other image-capturing solutions. Our technology is chassis-agnostic.

Availability and customers

When will Vinsa’s computer vision be available with Boston Dynamics’ Spot quadruped robot?

Perry: It is available now and already operating with customers in the electric utility and manufacturing spaces.

Will Vinsa and Boston Dynamics’ joint solution be offered for sale or as a service?

Perry: Customers can contact our team, and we can walk them through getting Spot and access to the Vinsa platform and analytics service.

Bruce: Vinsa’s Intelligence Models and Boston Dynamics’ Spot are both available for monthly leasing.

Do you have customers in the pipeline? What sectors are they in?

Perry: Yes. Ultimately, the value of Spot is to make collecting large volumes of diverse data seamless. Incorporating Vinsa’s computer vision models makes this data actionable by our customers.

Bruce: Current applications with potential customers are in discussion in oil and gas, manufacturing, industrial, retail, and hospitality.

The post Vinsa to build visual AI workflows with Boston Dynamics’ Spot quadruped robot appeared first on The Robot Report.


Source: therobotreport

Leave a Reply