Scale AI > Case Studies > Nuro Enhances Autonomous Vehicle Safety with Nucleus Object Autotag

Nuro Enhances Autonomous Vehicle Safety with Nucleus Object Autotag

Scale AI Logo
Technology Category
  • Analytics & Modeling - Machine Learning
  • Cybersecurity & Privacy - Identity & Authentication Management
Applicable Industries
  • Education
  • Transportation
Applicable Functions
  • Logistics & Transportation
Use Cases
  • Tamper Detection
  • Virtual Training
Services
  • Training
About The Customer
Nuro is a robotics company based in Mountain View, CA, that aims to improve everyday life through robotics. The company's custom electric autonomous vehicles are designed to deliver the things people need, from produce to prescriptions, right to their homes. Nuro's R2 robot was the first completely autonomous, zero-occupant on-road vehicle approved for commercial delivery by the U.S. Department of Transportation. The company has partnered with brands like Domino's and FedEx and has piloted autonomous, eco-friendly local delivery for communities in Texas, Arizona, and California. Nuro's approach to building a custom-goods only vehicle allows it to prioritize safety innovations for vulnerable road users such as bikers and pedestrians.
The Challenge
Nuro, a robotics company specializing in autonomous vehicles for delivery services, faced a significant challenge in identifying infrequent but meaningful scenarios in their training data. The company's autonomous vehicles, designed to deliver goods from produce to prescriptions, needed to be able to identify and respond to a variety of obstacles, including pedestrians in unusual postures, animals, occluded and backlit pedestrians, and infrequently encountered vehicles such as excavators. However, these labels were not present in the ground truth of their training data. The company's internal tool was only able to identify a limited number of these scenarios, falling short of the thousands of images that needed to be identified and labeled for comprehensive training of their autonomous vehicles.
The Solution
Nuro turned to Nucleus, a tool that enables the company's Perception Team to collaborate and identify safety-critical edge cases in their training data. Nuro aimed to ingest a substantial portion of its training dataset into Nucleus to identify these rare edge cases from a massive collected dataset with minimal effort or intervention from its machine learning team. Nucleus' Object Autotag and data curation workflows allowed machine learning engineers to use a consistent platform for modeling tasks. Autotag enabled users to select unlabeled images of a certain category, and then using an internal model and its feature vectors, provide a set of similar images, suitable to be sent out for labeling with new class labels. This allowed the team to incorporate newly identified edge cases and re-train a new model in just a few days rather than over a week.
Operational Impact
  • With the help of Nucleus, Nuro has been able to tackle the ever-growing long tail of rare scenarios in their training data. The company's Perception Team now maintains an impressive dataset at a scale of over 500 million images. By holding the amount of labeled data fixed for supervised learning, they've been able to handle more and more edge cases, such as avoiding collisions with large birds. Nuro intends to continue to scale up the size of its dataset that it curates with Nucleus, preparing the perception team to target any new edge cases that may appear. Future training data will likely include changing environmental conditions and identifying actions, such as the hand-wave of a pedestrian. This has allowed Nuro to improve the safety and efficiency of their automated deliveries with fewer human interventions required.
Quantitative Benefit
  • Nuro was able to identify 1000+ images of pedestrians in unusual postures with Nucleus, compared to ~60 with their internal tool.
  • Nucleus helped Nuro identify 400+ images of animal cases, compared to ~50 with their internal tool.
  • With Nucleus, Nuro was able to identify 500+ images of occluded and backlit pedestrians, compared to 10-20 with their internal tool.

Case Study missing?

Start adding your own!

Register with your work email and create a new case study profile for your business.

Add New Record

Related Case Studies.

Contact us

Let's talk!
* Required
* Required
* Required
* Invalid email address
By submitting this form, you agree that IoT ONE may contact you with insights and marketing messaging.
No thanks, I don't want to receive any marketing emails from IoT ONE.
Submit

Thank you for your message!
We will contact you soon.