A Better Match for Drivers and Riders: Reinforcement Learning at Lyft

To better match drivers to riders in our ridesharing application, we revised Lyft's core matching algorithm. We use a novel online reinforcement learning approach that estimates the future earnings of drivers in real time and use this information to find more efficient matches. This change was the first documented implementation of a ridesharing matching algorithm that can learn and improve in real time. We evaluated the new approach during weeks of switchback experimentation in most Lyft markets, and estimated how it benefited drivers, riders, and the platform. In particular, it enabled our drivers to serve millions of additional riders each year, leading to more than $30 million per year in incremental revenue. Lyft rolled out the algorithm globally in 2021.

PAPRAS: Plug-And-Play Robotic Arm System

This paper presents a novel robotic arm system, named PAPRAS (Plug-And-Play Robotic Arm System). PAPRAS consists of a portable robotic arm(s), docking mount(s), and software architecture including a control system. By analyzing the target task spaces at home, the dimensions and configuration of PAPRAS are determined. PAPRAS's arm is light (less than 6kg) with an optimized 3D-printed structure, and it has a high payload (3kg) as a human-arm-sized manipulator. A locking mechanism is embedded in the structure for better portability and the 3D-printed docking mount can be installed easily. PAPRAS's software architecture is developed on an open-source framework and optimized for low-latency multiagent-based distributed manipulator control. A process to create new demonstrations is presented to show PAPRAS's ease of use and efficiency. In the paper, simulations and hardware experiments are presented in various demonstrations, including sink-to-dishwasher manipulation, coffee making, mobile manipulation on a quadruped, and suit-up demo to validate the hardware and software design.