Wayve and Alex Kendall: Revolutionizing Embodied AI

Wayve and Innovation in Autonomous Driving

In the world of technology and industry, transformations are often driven by individuals with diverse backgrounds. Some explore uncharted territories, others build companies capable of realizing these ideas, and still others change the rules of the game or provoke lasting change. The series Architects of the Future examines four ways to wield the power of innovation, highlighting the specific role of leaders in the dynamics of their time. The first installment of this series is dedicated to Alex Kendall, the founder of Wayve.

An Exceptional Journey

Alex Kendall caught attention in 2018 when he won a pitch competition at WebSummit. Five years later, he finds himself speaking in the grand auditorium of the engineering department at Cambridge, discussing the concept of Embodied AI. At the helm of Wayve, a prominent company in the field of autonomous driving, Kendall recently raised €1.2 billion and is considering an initial public offering. For him, the first wave of artificial intelligence was just the beginning. The next step is to develop machines capable of perceiving their environment, understanding scenes, and acting in the real world.

A Childhood Marked by Exploration

The term "boundaries" frequently appears in Alex Kendall's discourse, whether in research, entrepreneurship, or his childhood in New Zealand. Growing up in this island nation was a formative experience, instilling values of perseverance and curiosity. Shortly after finishing high school, he embarked on building a drone, a project he considers a turning point. He realized that some technologies emerge at the intersection of multiple disciplines, and that robotics and autonomous systems require a synergy between artificial intelligence, electronic engineering, and mechanics.

The Silicon Valley Experience

Before joining Cambridge, Kendall spent time in Silicon Valley, where he participated in the development of a consumer drone. This experience confirmed for him that technology must be grounded in practical uses. His time in California provided him with insights into tech entrepreneurship, characterized by a culture of rapid experimentation and acceptance of failure.

Cambridge: A Laboratory for New Ideas

At Cambridge, Kendall focused on machine learning and computer vision, essential fields for machines operating in the physical world. During his PhD, he contributed to the development of SegNet, a deep learning system designed to interpret images captured by a camera. Kendall is convinced that machines can learn to understand the visual world and make decisions based on it. This intuition aligns with research on agents learning in simulated environments, notably at DeepMind.

The Wayve Challenge

In 2017, Alex Kendall defended his thesis and co-founded Wayve with Amar Shah. The initial prototypes were developed under modest conditions, using a small electric Renault Twizy equipped with cameras. This vehicle began to move without human intervention, without lidar, without heavy infrastructure, and without reliance on mapping. The idea behind Wayve is to teach a car to drive using data, rather than explicitly coding all the rules. Kendall explains that they built an end-to-end deep learning system from the start.

A Bold Approach

At the time, this approach faced significant skepticism, as most industry players favored modular architectures. Kendall and Shah sought to train a system capable of observing its environment and directly deducing a driving decision. The early years were challenging, but the technology progressed, with prototypes learning to navigate increasingly complex environments. In 2019, Wayve reached a pivotal milestone with a vehicle capable of navigating unknown roads.

From Startup to a Fundamental Model

Over time, Wayve has evolved into a structured technology company with a broader goal than just autonomous driving, aiming to build an embodied AI foundation model. The strategy is to train a model capable of piloting different vehicles and learning from data sourced from multiple fleets. Unlike Tesla, which improves its system based on its own fleet, Wayve aims to sell its model to various automotive OEMs to aggregate more diverse data.