Real-time Performance

Time and tide wait for no one.

In the real world, people and objects will move around the environment, the world will change, and a robot cannot stay stationary and wait, scratching its head to think of what to do. It is essential for critical robotics algorithms that guarantee safety and liveness (i.e., will be able to reach the goal) to be able to react immediately to changing circumstances. Autonomous cars must make decisions that determine the safety of their passengers and the world around them, mobile manipulators working in a factory or a warehouse may need to dodge unseen people and obstacles, and surgical robots must maintain patient safety at all times.

Algorithms that provide real-time performance can come through theoretical properties (guaranteeing a solution in a fixed amount of computation) or software/hardware engineering (using implementation or platform-specific features to accelerate algorithms).


2024

  1. Collision-Affording Point Trees: SIMD-Amenable Nearest Neighbors for Fast Collision Checking
    In Robotics: Science and Systems
  2. Motions in Microseconds via Vectorized Sampling-Based Planning
    In IEEE International Conference on Robotics and Automation
  3. Workshop
    Dynamic Motion Planning from Perception via Accelerated Point Cloud Collision Checking
    In IEEE ICRA 2024 Workshop—Agile Robotics: From Perception to Dynamic Action

2021

  1. HyperPlan: A Framework for Motion Planning Algorithm Selection and Parameter Optimization
    In IEEE/RSJ International Conference on Intelligent Robots and Systems