Company applies deep learning to create scalable AI approach for ADAS L2/L3 through L4
REDWOOD CITY, California., December 28, 2023 /PRNewswire/ — Helm.ai, a provider of next-generation AI software for autonomous driving and robotics automation, today announced Deep Neural Network (DNN)-based base models for prediction behavioral and decision-making within the company’s AI software. battery for high-end autonomous driving ADAS L2/L3 and L4.
The company trained DNN base models to make predictions about vehicle and pedestrian behavior in complex urban scenarios, as well as to predict the path an autonomous vehicle would take in these situations, which are essential ingredients of the capabilities personal decision-making. -drive cars. Helm.ai leveraged its industry-validated full-scene semantic segmentation and panoramic 3D sensing system as a base representation to enable training intent prediction and path planning capabilities. Additionally, base models are trained using the company’s proprietary Deep Teaching technology to achieve broad predictive capability in a scalable manner.
Helm.ai’s technology learns directly from real driving data and uses the company’s highly accurate and temporally stable perception system to capture information about complex vehicle and pedestrian behaviors and the driving environment surrounding environment, leading to DNNs that automatically learn the subtle but important aspects of urban driving. . The core models that power Helm.ai’s intent and path prediction gather data from a series of observed images and generate predicted video sequences that represent the most likely possible outcomes of what will happen next. The models also provide a predicted path for the autonomous vehicle that is consistent with the intent prediction. Intent prediction and trajectory prediction capabilities are essential for planning the safest optimal action of the autonomous vehicle.
Importantly, Helm DNN base models for intention prediction and path planning are trained under the highly scalable Deep Teaching paradigm, enabling unsupervised learning on complex urban driving scenarios directly from real driving data. This approach avoids heavy physics-based simulators and hand-coded rules, which are insufficient to capture the full complexity of real-world driving. In particular, the Helm.ai development and validation pipeline, while optimized for high-end ADAS L2/L3 mass production software, can also be directly applied to fully standalone L4 applications. Additionally, Helm.ai’s scalable AI approach easily generalizes to areas of robotics beyond autonomous vehicles.
Helm.ai is developing an AI-driven approach to autonomous driving, designed to scale seamlessly from high-end ADAS L2/L3 mass production programs to large-scale L4 deployments. The company’s software-only platform is hardware-agnostic and vision-focused, solving the critical problem of vision perception while also integrating sensor fusion between vision and radar/lidar where necessary. The technology advancements announced today accelerate the value of Helm.ai’s software offering by paving the way for the scalable development and validation of AI-based intent prediction and path planning software for vehicles autonomous.
“At Helm.ai, we are pioneering a highly scalable AI approach that simultaneously addresses mass production of high-end L2/L3 ADAS and large-scale L4 deployments in the same framework,” said Helm.ai. CEO of Helm.ai. Vladislav Voroninsky.
“Perception is the first essential component of any autonomous driving stack. The more complete and stable a perception system is over time, the easier it is to create downstream prediction capabilities, which is particularly critical for complex urban environments. panoramic view urban perception system and Deep Teaching training technology, we trained DNN base models for intention prediction and path planning to learn directly from real driving data, enabling them to understand a wide variety of urban driving scenarios and the intricacies of human behavior without the need for traditional physics-based simulators or hand-coded rules.
Helm.ai closed a $55 million Series C funding round in progress August 2023. The round was led by Freeman Group and includes investments from venture capital firms ACVC Partners and Amplo as well as strategic investments from Honda Motor, Goodyear Ventures and Sungwoo Hitech. This funding brings the total amount raised by Helm.ai to $102 million.
About Helm.ai
Helm.ai develops the next generation of AI software for high-end ADAS, L4 autonomous driving and robotics. Founded in November 2016 In Menlo Park, California, the company has reimagined how AI software is designed to make truly scalable autonomous driving a reality. For more information about Helm.ai, including its products, SDK, and open career opportunities, visit https://www.helm.ai/ or find Helm.ai at LinkedIn.
Photo – https://mma.prnewswire.com/media/2307805/Intent___Path_Prediction_Top_3.mp4
Logo – https://www.times24h.com/wp-content/uploads/2023/12/Helmai-Announces-DNN-Core-Models-for-Intent-Prediction-and-Path.jpg
Show original content:https://www.prnewswire.com/apac/news-releases/helmai-announces-dnn-foundation-models-for-intent-prediction-and-path-planning-302022566.html
SOURCE Helm.ai
