top of page

Students at Harvey Mudd College Successfully Trained Syntiant's NDP101 for Sensor App in Smartwatch

Deep Learning App Detects Motion and Gesture Recognition; Expands Uses Beyond Voice; Demonstrates High Accuracy, Reduced Latency and Ultra-Low Power Consumption

Students from Harvey Mudd College successfully designed and trained the Syntiant® NDP101™ Neural Decision Processor™ (NDP) to detect and classify hand motions for a wearable device.

Students from Harvey Mudd College's Clinic Program are flanked by Syntiant's VP of Hardware Dave Garrett (left) and Prof. David Harris (right).

As part of Harvey Mudd College’s Clinic Program, students collaborated with Syntiant's engineering team to demonstrate the versatility and efficiency of the company's NDP by designing a battery-powered application that receives live data from sensors and uses a neural network to detect significant events.

“The students at Harvey Mudd College did an outstanding job on the project, further validating that our ultra-low-power silicon can provide greater efficiency for sensor applications at the edge,” David Garrett, VP of hardware engineering at Syntiant, who led the collaboration. “As we continue to focus on addressing global demand for always-on voice control, we believe that the sky is the limit for motion and gesture sensing in smart devices that are highly accurate and consume significantly less power than traditional MCU solutions.”

The students collected more than 60,000 motions and gestures.

Utilizing Syntiant’s NDP9101 development board and deep learning training development kit, the students collected and trained more than 60,000 motions and gestures, from standing still to watch checking and wrist rotation (supination and pronation).

The team of undergraduates achieved 94 percent motion accuracy on the neural network.

“This was an outstanding project,” said Dr. David Harris, Ph.D., professor of engineering design at Harvey Mudd College, who served as the project advisor. “The students had an opportunity to explore machine learning, sensors and low-power circuit design. They learned an enormous amount and delivered a system that Syntiant is already demonstrating to potential customers. Syntiant was a terrific sponsor, providing the team with a challenging open-ended problem with clear business value, and supported the team with in-house expertise, while giving the students freedom to define their own applications and technical approaches."

Students from Harvey Mudd College's Clinic Program that worked on the Syntiant project.

Founded as an innovation in engineering education in 1963, Harvey Mudd College's Clinic Program has been copied by institutions worldwide. Under the guidance of a faculty advisor and a company liaison, students work in teams of four or five to develop solutions to unsolved problems presented by sponsoring organizations. All intellectual property generated through Clinic belongs to the sponsor.

Custom built to run AI workloads, Syntiant’s architecture provides 100x more efficiency and 10x the throughput of the typical microcontroller unit (MCU) solution with ultra-low power consumption. The result is an inference solution capable of intelligent processing at the edge for a wide variety of voice and sensor applications in battery-powered smart devices, free from a cloud connection, ensuring privacy and security.

Click on the link below to download the student presentation detailing the project’s findings.

Harvey Mudd College Clinic Program - Syn
Download • 6.22MB

For more information, contact Syntiant at 

Recent Posts

See All
bottom of page