Robotics
Overhaul
A Silicon Valley start-up built a game-changing multi-purpose robot powered by computer vision and IoT sensors, but before it could move from beta to launch it needed a major boost in processing speed, ease of use, and reliability.
Three Critical Problems
The IoRT (Internet of Robotic Things) product had tremendous potential, receiving interest from government, military, academic and corporate research labs, manufacturing facilities, and classrooms. Reaching that potential required solving these key problems.
Slow Computer Vision
The system took 1-2 minutes to analyze each image. It was often tripped up by variations in lighting conditions and failed to recognize translucent or transparent materials.
Disjointed Control
The robot was controlled by two software programs running on separate devices, requiring user input from both the robot's display screen and the user's computer.
Confusing UX
The user interface appeared outdated and the user experience was not intuitive. Beta users struggled to understand how to use the product.
Complete System Transformation
A New Way of Seeing
We replaced the standard computer vision system with deep learning visual processing. Training a neural network requires thousands or millions of images - something our client didn't have.
AI dev's proprietary data augmentation systems turned their few thousand images into over a million, successfully training the network for implementation.
- Works in diverse lighting conditions
- Detects translucent materials
- Recognizes transparent objects
- 20x faster processing
Simplified & Centralized Control
Instead of working within the confines of the current framework, we saw the computer vision upgrade as an opportunity to reimagine the entire user experience from scratch.
By automating multiple user input processes, we centralized all controls into one program, allowing full control of the robot's rich functionality from a single interface.
- Single device operation
- Automated input processes
- Full functionality access
- Streamlined workflow
Instructions Not Included
We tore down the outdated and confusing interface to build a new one from the ground up - beautifully designed and easy to use.
Our goal was to re-invent user interaction from top to bottom, making it so simple that users understand intuitively with no written instructions. New users glide through the process simply by following visual cues and on-screen prompts.
- Intuitive visual cues
- On-screen guidance
- Zero learning curve
- Modern aesthetic
The Transformation
We took a complex robotic system with a 100+ page manual and simplified it to the point where a manual was no longer necessary.