Autonomous trucking pioneer Kodiak Robotics is utilizing Embedl’s cutting-edge edge AI tools to enhance the deployment of generative AI models across multiple hardware platforms. This collaboration enables Kodiak to optimize performance and scalability, reinforcing its position as a leader in the autonomous vehicle (AV) industry.
Key Highlights:
- Edge AI Innovation: Kodiak integrates advanced neural networks and multimodal sensor fusion, combining camera and lidar data for precise real-time decision-making.
- Generative AI Application: Kodiak’s use of Generative AI (GenAI) handles complex and novel driving scenarios, ensuring operational reliability across diverse terrains.
- Embedl’s Optimization Impact: Embedl’s hardware abstraction technology enables seamless model deployment across NVIDIA and Ambarella hardware, significantly improving inference speeds.
- Efficiency Gains: Kodiak engineers leveraged Embedl’s SDK to transition vision and multimodal GenAI models between platforms in weeks, achieving optimized performance.
At the core of Kodiak’s autonomous trucking system lies edge AI models, enabling real-time decision-making on embedded hardware within the vehicle. These models process immense data streams to predict and adapt to dynamic driving conditions. Kodiak’s system uniquely utilizes Generative AI to handle diverse and challenging environments, such as snowy terrains or dense forests, bolstering the autonomy of its fleet.
Embedl plays a critical role in optimizing Kodiak’s neural networks for deployment on AV hardware. By automating the optimization process, Embedl ensures that Kodiak’s AI models perform efficiently on embedded hardware accelerators, such as NVIDIA and Ambarella platforms. This not only enhances Kodiak’s software stack but also allows for simultaneous support of multiple hardware platforms, streamlining development efforts.
Shubham Shrivastava, Head of Machine Learning at Kodiak Robotics, emphasized the value of Embedl’s tools: “Having the ability to deeply inspect the cognitive blocks of our AI models, perform hardware-aware optimization, benchmark various layers, and deploy models through seamless hardware abstraction is truly game-changing.”
The partnership highlights how cutting-edge tools like Embedl’s SDK can transform the scalability and performance of AI-driven autonomous vehicles, propelling innovation in the trucking and logistics industries. With faster inference speeds and a robust system optimized for diverse environments, Kodiak is paving the way for the future of autonomous transportation.
