The robot car coming to London

Key Concepts

• Whimo: A driverless car developed for operation in London.
• LiDAR: Light Detection and Ranging – a remote sensing technology using laser light to create a 3D map of the surroundings.
• Autonomous Driving: The capability of a vehicle to navigate and operate without human input.
• On-Device Compute: Processing data and making decisions locally within the vehicle, rather than relying on cloud connectivity.
• Phased Approach: A gradual rollout strategy involving data collection, training, and testing before public rider access.

Introduction of the Whimo Driverless Car

The video introduces the Whimo, a driverless car slated for public operation in London towards the end of the year. Currently, approximately two dozen Whimo vehicles are operating in London, but are being driven by humans for data collection purposes. The intention is to allow public ridership later in the year, pending government approvals, and in partnership with both local and national government bodies.

Sensor Suite and Technological Components

The Whimo utilizes a comprehensive suite of sensors to perceive its environment. These include:

• LiDAR (Light Detection and Ranging): Located on the roof, LiDAR provides 360-degree vision, enabling operation in low-light conditions, surpassing human capabilities in nighttime visibility.
• Cameras: Functioning similarly to human eyes, cameras provide visual data. The system’s capabilities extend beyond human vision. Additional cameras are positioned around the vehicle for comprehensive coverage.
• LiDAR (Rear-Facing): A spinning LiDAR sensor is located at the rear of the vehicle, providing a view of the surroundings behind the car.
• Microphones: These act as “ears,” detecting sounds like sirens, enhancing situational awareness.

The core of the Whimo’s functionality lies within the “incredibly powerful chips” housed inside the vehicle. These chips facilitate “on-device compute,” meaning all processing and decision-making occur locally, without relying on cloud connectivity. This is crucial for rapid response times, particularly in critical situations like a pedestrian entering the roadway. The speaker emphasizes that relying on cloud processing would introduce unacceptable delays.

Addressing London’s Unique Driving Challenges

London presents specific challenges for autonomous vehicles due to its complex urban environment. However, the developers are leveraging data collected from 127 million miles of fully autonomous driving in the US to address these challenges. Specifically, the data indicates a significant safety improvement:

• Pedestrian Crash Reduction: The Whimo system has been involved in 92% fewer crashes with pedestrians compared to human drivers. This suggests the system is capable of safely navigating around unpredictable pedestrian behavior, including “jaywalking.”

The company is employing a “phased approach” to deployment. This involves:

1. Data Collection: Gathering data on London’s specific road conditions and pedestrian patterns.
2. City Learning: Utilizing the collected data to train the autonomous driving system.
3. Driver Training: Refining the system’s performance based on the learned data.

This phased approach aims to ensure the system is fully prepared and safe for public ridership. The geography of London is identified as the biggest difference compared to testing environments.

Importance of On-Device Processing

A key argument presented is the necessity of “on-device compute.” The speaker states, “It doesn't work if you're like going up to the cloud to say, 'Oh, what should I do about this?' when a pedestrian runs into the road.” This highlights the critical need for real-time processing and decision-making capabilities within the vehicle itself, as latency introduced by cloud communication could have severe consequences.

Notable Quote

“And so actually the jaywalkers, the pedestrians, like that is something that we know we can behave very very safely around.” – Speaker, referencing the data collected from US autonomous driving experience.

Synthesis

The Whimo represents a significant step towards the implementation of driverless technology in a complex urban environment like London. The vehicle’s advanced sensor suite, coupled with powerful on-device processing, aims to overcome the challenges posed by unpredictable pedestrian behavior and intricate city layouts. The phased deployment strategy, driven by extensive data collection and analysis, underscores a commitment to safety and responsible innovation. The core takeaway is that the success of autonomous driving hinges on the ability to process information and react in real-time, locally within the vehicle, rather than relying on external cloud connectivity.