Why Self-Driving Trucks Are So Hard! Data, Sensors, AI & Real-World Driving
What does it actually take to build self-driving trucks that can interpret the real world and react faster than humans?
In this episode of An Hour of Innovation podcast, Vit Lyoshin speaks with Achyut Boggaram about what it really takes to build autonomous trucking systems that operate safely in the real world. Rather than focusing solely on AI breakthroughs, the conversation highlights a deeper reality: deploying self-driving technology is fundamentally a systems engineering challenge.
Achyut explains how autonomous trucks rely on a combination of sensors, data pipelines, machine learning models, and traditional rule-based systems working together. At the core is not just intelligence, but reliability, especially in rare and unpredictable situations. These long-tail scenarios, which may occur infrequently but carry high risk, are the primary focus of modern autonomous system development.
The discussion explores how safety is embedded into every layer of the system, from motion planning and decision-making to fail-safe mechanisms that ensure the vehicle can default to minimal-risk behavior when uncertainty arises. Achyut also shares how the industry has evolved from research labs into real-world deployment, with companies now optimizing for scale, data collection, and operational performance.
Beyond the technical perspective, the episode touches on the broader impact of autonomous trucking on logistics and supply chains. From reducing risks in critical situations, such as pandemic response, to addressing labor shortages in long-haul trucking, autonomous systems have the potential to reshape how goods are transported.
This conversation offers a grounded, practical view of AI in production, showing that the future of autonomy depends not just on smarter algorithms, but on building systems that can handle the complexity and unpredictability of the real world.
Achyut Boggaram is an AI and machine learning engineer working on autonomous driving technology at Torc Robotics. His work focuses on building the machine learning infrastructure and data pipelines that power self-driving truck models at scale. After leading ML platform development, he moved into applied research, where he contributes directly to developing frontier AI models for autonomous vehicles. His experience offers a rare inside perspective on how modern robotics engineering and AI systems come together to power real-world autonomous driving software.
Takeaways
- A single 20-minute autonomous truck test run can generate about 100 terabytes of raw sensor data, showing how data-intensive self-driving systems really are.
- Autonomous trucks rely on sensor fusion from cameras, lidar, radar, GPS, and IMU sensors to build a real-time understanding of the road.
- Self-driving systems are structured in layers: perception models understand the environment, while planning and behavior models decide what the vehicle should do next.
- Machine learning models must generalize from training examples.
- When the AI becomes uncertain, the system can execute a minimal-risk maneuver, such as slowing down and pulling off the road.
- Training autonomous vehicle models requires diverse real-world data across conditions like night driving, fog, rain, and heavy traffic.
- Engineers often use synthetic data and neural rendering to simulate rare scenarios that are difficult or dangerous to capture in real life.
- Autonomous driving systems must be designed to resist adversarial attacks, where small visual changes can trick AI into misinterpreting road signs.
- AI-powered perception systems can sometimes detect objects hundreds of meters away and even see through fog.
- Reinforcement learning and large-scale simulation allow engineers to train driving behaviors without putting humans at risk on real roads.
- The biggest barrier to widespread deployment is the “long tail” of rare driving scenarios, where unpredictable real-world situations challenge even the best AI systems
Timestamps
00:00 Introduction
03:46 Understanding Autonomous Vehicles
06:56 Components of Autonomous Vehicle Technology
14:41 Generalization in Machine Learning Models
19:11 Data Collection and Processing for Training Models
24:47 Adversarial Attacks and Model Robustness
27:00 Advanced Sensor Technologies for Hazardous Conditions
29:41 Reinforcement Learning in Autonomous Vehicles
32:45 Challenges in Self-Driving Technology Adoption
37:13 Future of Logistics and Autonomous Vehicles
40:37 Innovation Q&A
Support the Podcast
To support our work, please check out our sponsors and get discounts: https://www.anhourofinnovation.com/sponsors/
For inquiries about sponsoring An Hour of Innovation, email iris@anhourofinnovation.com
Connect with Achyut
- Website: https://torc.ai/
- LinkedIn: https://www.linkedin.com/in/achyutsarma/
Connect with Vit
- Substuck: https://anhourofinnovation.substack.com/
- LinkedIn: https://www.linkedin.com/in/vit-lyoshin/
Episode References
Daimler Truck
https://www.daimlertruck.com
A global commercial vehicle manufacturer (part of Mercedes-Benz Group) that builds trucks used for autonomous driving platforms.
Waymo
https://waymo.com
An autonomous vehicle company known for deploying robotaxis and using lidar-based perception systems.
Tesla
https://www.tesla.com
A company developing full self-driving (FSD) systems using primarily camera-based perception and large-scale real-world data.
Reinforcement Learning
https://en.wikipedia.org/wiki/Reinforcement_learning
A machine learning method where systems learn by receiving feedback (rewards or penalties) from their actions.
AlphaGo
https://en.wikipedia.org/wiki/AlphaGo
An AI system developed by DeepMind that mastered the game of Go using reinforcement learning techniques.
ChatGPT
https://chat.openai.com
A conversational AI model trained using reinforcement learning and human feedback to generate natural language responses.
Adversarial Attacks
https://en.wikipedia.org/wiki/Adversarial_machine_learning
Techniques used to trick AI systems by subtly altering inputs, causing incorrect predictions or behaviors.
