Boston Scientific recently provided an amusing update to its quadrupedal robot, Spot through an upgrade powered by ChatGPT.
Matt Klingensmith and the software team from Boston Scientific were able to give Spot a voice that is able to listen and respond to input from humans.
Using generative AI from ChatGPT combined with video cameras serving as cameras, and a text-to-voice feature the robot can see, listen, and respond to visitors’ questions.
The team provided the robots with a variety of personalities including an English butler, a 1920s archeologist, a teenager, and a time traveler from the time of Shakespeare.
Klingensmith provided the robots with a brief script that describes the rooms at the Boston Scientific labs and now the robots can serve as a tour guide to the site.
The robot starts by asking the names of the people on the tour and then refers to them by name throughout the interaction.
The developers even programmed the robot’s claw to move when the robot talks emulating talking.
By combining input from the cameras mounted on the robot’s “head” it is able to categorize items that are around it and incorporate them into the conversation.
The robots are capable of surprising the developers with their answers.
When the robot was asked who their parents were, the robot took the developer to a museum room that featured the first prototypes of the Spot robot.
While the project is relatively simple, Klingensmith believes that “a world in which robots can generally understand what you say and turn that into useful action is probably not that far off.”