Waabi's Raquel Urtasun on Launching an AV Technology Startup

Waabi Emerges: A New Player in Autonomous Trucking

Raquel Urtasun, previously the chief scientist at Uber ATG, has launched Waabi, a startup dedicated to developing autonomous vehicle technology.

The company officially exited stealth mode last week, announcing its focus on the trucking sector.

Series A Funding and Leadership

Waabi, headquartered in Toronto, secured $83.5 million in Series A funding. This substantial investment was spearheaded by Khosla Ventures.

Urtasun currently serves as both the founder and CEO of Waabi, guiding the company’s strategic direction.

Insights from Mobility 2021

Urtasun participated in Mobility 2021, where she discussed her new venture and the broader landscape of autonomous vehicle development.

Her presentation addressed the significant hurdles currently confronting the self-driving vehicle industry.

A Novel AI Approach

A key focus of Urtasun’s discussion was how her unique approach to AI can accelerate the process of bringing autonomous vehicles to market.

She detailed how this innovative methodology aims to overcome existing limitations and facilitate the commercialization of AVs.

The core of Waabi’s strategy centers around leveraging advanced artificial intelligence to enhance the safety and efficiency of autonomous trucking operations.

The Genesis of Waabi: Urtasun's Entrepreneurial Leap

Recognized as a leading figure in the field of Artificial Intelligence, Raquel Urtasun previously directed research and development as Chief Scientist at Uber ATG. Following the acquisition of Uber ATG by Aurora in December, a new venture emerged just six months later: Waabi.

The core mission of Waabi centers on leveraging an AI-first methodology to overcome the challenges inherent in developing self-driving technology.

Context: The Transition from Uber ATG to Waabi

Prior to the formation of Waabi, significant industry events unfolded. Discussions regarding the potential sale of Uber ATG’s self-driving unit to Aurora were widely reported.

Ultimately, Uber finalized the sale of Uber ATG, a transaction that substantially increased Aurora’s valuation, reaching $10 billion.

• Reports detailed Uber’s negotiations to divest its ATG self-driving division to Aurora.
• The completion of the sale saw Uber transfer its Uber ATG unit, significantly boosting Aurora’s overall worth.

Challenges Hindering Widespread Adoption of Commercial AV Systems

Full autonomy in the autonomous vehicle sector remains an elusive goal. While companies such as Cruise and Waymo have initiated limited deployments of robotaxi services, the underlying technology currently lacks the robustness required for broader scalability.

Current Deployments and Their Limitations

Despite progress, the technology powering these systems isn't yet mature enough to support large-scale operations. Existing deployments are confined to specific, carefully mapped areas.

The complexities of real-world driving conditions present significant hurdles. Unexpected events and unpredictable pedestrian or driver behavior require sophisticated decision-making capabilities that are still under development.

Recent Developments in the Industry

• Waymo has begun offering driverless ride-hailing services to the general public.
• A shift in terminology is occurring, with Waymo moving away from the term ‘self-driving,’ although this change isn't universally accepted within the industry.
• Cruise is now authorized to provide driverless passenger transport within California.

These developments represent incremental steps forward, but they also highlight the ongoing challenges. Expanding these services beyond controlled environments necessitates substantial advancements in sensor technology, artificial intelligence, and safety protocols.

Scalability is a key concern. Replicating the performance of these systems across diverse geographical locations and varying weather conditions demands significant engineering effort and ongoing refinement.

Furthermore, regulatory frameworks and public perception play a crucial role. Establishing clear guidelines and building public trust are essential for the successful integration of autonomous vehicles into mainstream transportation.

Balancing Deep Neural Networks and Rule-Based AI in Autonomous Systems

Raquel Urtasun’s methodology for achieving autonomy integrates both deep neural networks and rule-based AI. This blended approach mirrors the conventional strategy employed in the development of self-driving vehicles.

A primary reason for this preference stems from the critical nature of human transportation. Developers often exhibit caution regarding the use of deep neural networks due to the inherent "black box" problem.

This issue arises because it can be difficult, if not impossible, to determine the precise reasoning behind a deep learning system’s actions, hindering verification and validation processes.

The Role of Rule-Based Systems

Rule-based AI offers a level of transparency that deep neural networks often lack. With rules, engineers can explicitly define the system’s behavior and understand the logic behind its decisions.

This predictability is particularly important in safety-critical applications like autonomous driving, where accountability and reliability are paramount.

Related Developments in Autonomous Vehicle Technology

• Mate Rimac’s focus is on the creation of electric robotaxis.
• Cruise has recently secured a $2 billion investment from Microsoft.

The integration of these two AI paradigms – deep learning and rule-based systems – represents a significant step towards building robust and trustworthy autonomous vehicles.

By leveraging the strengths of each approach, developers aim to create systems that are both intelligent and demonstrably safe for public use.

Waabi's Innovative Approach: Minimizing Road Testing Through Advanced Simulation

The development of self-driving technology has traditionally involved extensive on-road testing to evaluate vehicle performance in genuine, everyday scenarios. This process, however, presents both logistical and regulatory challenges for many companies striving to advance and expand their autonomous systems.

Waabi is adopting a different strategy, placing significant emphasis on simulating real-world road conditions as a crucial component of their testing procedures.

The Traditional Road Testing Paradigm

Historically, deploying vehicles on public roads has been integral to achieving higher levels of autonomous functionality. Successfully navigating both traffic laws and obtaining necessary permits has been a fundamental requirement for startups aiming to develop and scale their self-driving capabilities.

Waabi’s Simulation-First Philosophy

Waabi’s approach centers around leveraging a sophisticated simulator. This allows for comprehensive testing of its vehicles in a wide range of virtual environments, reducing the need for extensive physical road testing.

• Driverless testing by Cruise has commenced in San Francisco.
• The CEO of Motional has indicated a future for autonomous technology within the logistics sector.

Simulation offers a controlled and efficient method for evaluating and refining autonomous systems before real-world deployment.

By prioritizing simulation, Waabi aims to accelerate the development process and address the complexities associated with traditional road testing.

Potential Conflicts Arising from Corporate Partnerships

Both Uber and Aurora hold investments in Waabi, yet Aurora is actively pursuing autonomous long-haul trucking – a strategic direction also adopted by Waabi with its self-driving technology. This overlap raises the question of potential competition or conflicts of interest for the developing company.

Urtasun posits that resolving this challenge necessitates cross-industry collaboration, acknowledging its inherent complexity.

Recent Developments in Autonomous Trucking

• A new pilot program will see Waymo and J.B. Hunt introduce autonomous trucks to Texas.
• Aurora is incorporating external expertise to enhance safety protocols and foster public confidence in driverless vehicle technology.
• Users can now access Waymo’s driverless taxi service directly through the Google Maps platform.

The autonomous vehicle sector is characterized by strategic alliances and overlapping interests.

Successfully navigating these relationships will be crucial for Waabi’s long-term success.

Collaboration, while beneficial, requires careful management to avoid competitive disadvantages.

The Impact of Industry Consolidation on Diverse Thinking in Autonomous Driving

The autonomous driving sector is experiencing both expansion and consolidation. Larger companies are increasingly acquiring innovative startups. While these newer companies often possess novel concepts, they frequently lack the financial capacity for widespread commercial implementation.

This trend raises concerns about potential limitations to innovation and the overall advancement of the field.

Recent Developments

Several recent events highlight this dynamic. For example, Cruise has recently finalized an agreement to introduce its robotaxi service in Dubai.

Furthermore, Uisee, an autonomous driving startup, has secured a $150 million investment from a Chinese state investor.

The complete discussion transcript is available for review here.

Related TechCrunch Mobility 2021 Sessions

Attendees of TechCrunch Mobility 2021 explored related topics in several sessions, including:

• AVs: Past, Present and Future – A comprehensive overview of autonomous vehicle development.
• The Rise of Robotaxis in China – Examining the growth and deployment of robotaxi services within China.
• From Concept to Commuter Car – and Beyond – Discussing the journey of autonomous vehicle technology from initial ideas to practical applications.

Explore all sessions from Justice 2021 by visiting this link.