[Paper Review] ZJUNlict Extended Team Description Paper for RoboCup 2019

Zheyuan Huang, Lingyun Chen+7 more|arXiv (Cornell University)|May 22, 2019

Robotic Path Planning Algorithms2 references4 citations

TL;DR

This paper presents ZJUNlict's advanced dribbler design and intelligent ball interception system for RoboCup 2019 Small Size League, featuring a 3-touch-point mechanical structure with bidirectional damping and screw-textured dribbling bar, enabling stable control at 8.5 m/s incoming ball speed. The team integrated Kalman filter-based vision loss compensation and a search-based interception prediction algorithm, reducing interception time by up to 40% and increasing ball possession through dynamic skill selection based on angular deviation.

ABSTRACT

For the Small Size League of RoboCup 2018, Team ZJUNLict has won the champion and therefore, this paper thoroughly described the devotion which ZJUNLict has devoted and the effort that ZJUNLict has contributed. There are three mean optimizations for the mechanical part which accounted for most of our incredible goals, they are "Touching Point Optimization", "Damping System Optimization", and "Dribbler Optimization". For the electrical part, we realized "Direct Torque Control", "Efficient Radio Communication Protocol" which will be credited for stabilizing the dribbler and a more secure communication between robots and the computer. Our software group contributed as much as our hardware group with the effort of "Vision Lost Compensation" to predict the movement by kalman filter, and "Interception Prediction Algorithm" to achieve some skills and improve our ball possession rate.

Motivation & Objective

• To overcome the limitations of passive 2-touch-point dribblers in high-speed, dynamic RoboCup Small Size League matches.
• To improve ball catching reliability and dribbling stability under aggressive maneuvers like turning and lateral movement.
• To enhance team coordination by predicting optimal interception points and assigning robots based on time-to-arrival rather than position.
• To increase ball possession rate through intelligent skill selection (chase, intercept, touch) based on angular deviation from target.

Proposed method

• Designed a 3-touch-point dribbler with dual support from carpet and chip shovel, enabling better force distribution and reduced bouncing.
• Implemented a bidirectional spring-damping system using taped side plates to absorb impact energy during baseplate collisions.
• Introduced a 3D-printed screw-textured dribbling bar to generate lateral forces during rotation, improving control during turns.
• Developed a vision loss compensation system using Kalman filtering to predict ball trajectory when tracking is lost.
• Proposed a search-based interception prediction algorithm that evaluates robot arrival time vs. ball position over discrete time steps.
• Implemented a dynamic skill selection framework that chooses 'Chase', 'Intercept', or 'Touch' based on angular deviation from target.

Experimental results

Research questions

• RQ1 Can a 3-touch-point mechanical design with bidirectional damping significantly improve ball catching performance compared to traditional 2-touch-point systems?
• RQ2 To what extent can a screw-textured dribbling bar enhance lateral control during robot rotation or lateral motion?
• RQ3 How effective is a time-based robot assignment strategy (using arrival time) for ball interception compared to position-based assignment?
• RQ4 Can vision loss compensation via Kalman filtering reduce the time to reacquire the ball after tracking failure?
• RQ5 What is the impact of dynamic skill selection on ball possession and goal-scoring efficiency in real-time RoboCup matches?

Key findings

• The new 3-touch-point dribbler with bidirectional damping successfully caught balls with an incoming speed of up to 8.5 m/s, a significant improvement over the 3 m/s limit of the traditional 2-touch-point design.
• The screw-textured dribbling bar enabled stable dribbling during rotation at 20°/s² acceleration, maintaining ball contact where smooth bars failed.
• The interception prediction algorithm reduced average interception time by up to 40% in high-speed scenarios, especially when ball speed exceeded 4 m/s.
• The marking skill, based on optimal interception prediction and circular defense zones, increased the likelihood of intercepting opponent passes by positioning within the critical radius around the predicted interception point.
• The FSM-based role matching system using time-to-arrival as a cost function improved ball possession by enabling faster, more accurate robot assignment to intercept the ball.
• The dynamic skill selection system (chase, intercept, touch) based on angular deviation from target improved execution accuracy and reduced unnecessary motion, contributing to faster transitions and higher scoring efficiency.