Billiards Robot: From Louis XI's Table to ABB Arms

A billiards robot sounds like a party trick, but building one runs straight into hard robotics problems. The machine has to find small, shiny balls on a table, choose a shot when every outcome is a little uncertain, and then hit with the same force every time. This post follows cue sports from royal tables in medieval France to the first computer and robot pool players. Then it looks at the present, where Chinese pool halls are testing AI tables that watch and coach players, and at where ABB arms such as GoFa and the IRB 1200 could fit. The last part covers what is still missing.

A snooker table with balls in play and cues resting on the cushion, 2022. Photo: Lee Vilenski / Wikimedia Commons (CC BY-SA 4.0)

From palace tables to pool halls with chalk and ivory

Billiards began outdoors. A recognisable form of the game was played on the ground in the 1340s, a little like croquet, before it moved onto tables. King Louis XI of France, who reigned from 1461 to 1483, had the first known indoor billiard table, and by the 1580s green cloth tables were recorded at Holyrood Palace in Scotland and in English noble houses. Players first pushed the balls with a mace, a stick with a wide foot. During the 1600s they started using its thin end for awkward shots, and by about 1800 the straight cue without a foot had taken over.

Carom billiards professionals including Welker Cochran, Kinrey Matsuyama and Willie Hoppe beside a table, 1 March 1927. Photo: National Photo Company Collection, Library of Congress via Wikimedia Commons (public domain)

Much of the game's later history is about materials. From 1627 until the early twentieth century, balls were mostly made of ivory, so they were expensive and never quite identical. John Wesley Hyatt invented celluloid in 1868 as the first workable ivory substitute, and William A. Spinks and William Hoskins invented modern cue tip chalk in 1897. In the United States, pool and billiards were hugely popular from the 1870s to the 1950s, the era of stars like Willie Hoppe, and the 1961 film The Hustler set off another boom. Through all of it, the shot depended on one person's eye and arm.

Willie Hoppe lining up a shot in 1927, when the game rested on the player's eye and arm. Photo: unknown photographer via Wikimedia Commons (public domain)

Computers came to the game through research labs. Pool makes a good robotics test because the physics is well known, yet every real shot carries small errors in aim, spin and speed. Computer pool tournaments let software players compete in simulation, and a program called CueCard won the 2008 contest. At Queen's University in Kingston, Ontario, Michael Greenspan's group built Deep Green, a gantry robot mounted on the ceiling with a wrist that held the cue. Ceiling cameras located the balls, and a camera on the wrist corrected the final aim. By 2008 the team said it played above amateur level.

Balls scattering on the break, the kind of uncertain outcome a pool robot has to plan around. Photo: No-w-ay with H. Caps / Wikimedia Commons (CC BY-SA 4.0)

AI tables, vision coaches and robot cue arms today

Most of the current work puts the intelligence into the table and leaves the cue in human hands. In September 2026, 36Kr described a prototype AI billiard table shown by Xiaotie, a Chinese operator of self-service pool halls with about 7,000 stores. A vision algorithm tracks the balls, four interactive screens are built into the table section, and an AI referee scores frames automatically. An AI coach looks at the previous shot and suggests where to hit the cue ball, which path to take and how much force to use. Sensors in the cue record each strike.

A player lines up a shot on a blue cloth table in a billiards hall in Pattaya, Thailand, 2025. Photo: PattayaPatrol / Wikimedia Commons (CC BY-SA 4.0)

Xiaotie is working in a crowded market. The same 36Kr report says China went from about 70,000 billiard venues in 2019 to more than 250,000 today, and some halls now charge 3 yuan an hour off peak, so a smart table is one way for a hall to stand out. The Queen's group tried an earlier form of the idea with ARPool, which projected suggested shots onto the cloth. Robots that actually hold the cue are still rare and mostly academic. They need a stiff, fast arm, a very precisely placed cue tip, and vision that keeps up when people walk around the table.

An evening game in a busy pool hall, the setting where smart tables now compete for players. Photo: PattayaPatrol / Wikimedia Commons (CC BY-SA 4.0)

ABB has not built a billiards robot, though its range covers most of the parts. A compact arm like the IRB 1200 can drive a cue along a straight line at high speed and repeat the stroke, which would suit a test rig for cues, tips or cloth. The GoFa CRB 15000 cobot is designed to work near people, so it could possibly set up practice positions or demonstrate a shot in a club. Nobody sells that today. Engineers would normally plan such a cell in RobotStudio and check the reach over a full size table before buying any hardware.

The new generation ABB IRB 1200, a compact arm launched in 2025 for fast, repeatable small part work. Photo: ABB

What a robot opponent at home would still need

In homes, the next few years probably look like the Xiaotie prototype: a camera above the table, a screen or projector, and software that scores, coaches and sends highlights to your phone. None of that needs a moving robot, so it can reach clubs and larger homes soon. A full robot opponent is much harder. A ceiling gantry like Deep Green needs a dedicated room, and a robot standing on the floor has to stay clear of the players and their cues. Cost is the biggest barrier, because a capable arm plus a good vision system costs far more than the table.

A player cues up on a pool table in a bar, the kind of club room where coaching cameras could arrive first. Photo: Derbeth / Wikimedia Commons (CC BY 2.0)

Robots may show up in the factories that make cue sports gear before they show up in pool rooms. Cues are turned and finished from wood and carbon, balls are cast and polished, and slate beds are cut and levelled to fine tolerance, and robots already do jobs like these in other industries. Arms could also test the products, striking thousands of identical shots to compare tips, measure how cushions rebound or see how cloth wears. ABB robots already do similar work in lab and quality cells, and you can find related examples in ABB robot use cases.

An ABB GoFa CRB 15000 cobot at a work table, an arm designed to work close to people. Photo: ABB

Several problems are still open. A cue moving fast at chest height can hurt someone, so any home or club robot needs speed limits and safety zones of the kind ABB's SafeMove provides. Players also care about spin and touch, and a robot that hits harder or straighter will not teach a beginner either one. And few people want to lose every game to a machine. The likeliest result is a mix of cameras and coaching software, with the occasional robot arm acting as a training partner that sets up drills and repeats shots on demand.

A GoFa cobot is guided by hand and stops when it senses contact, the kind of safety a club robot would need. Photo: ABB

Innovation

AI solutions for effortless ABB robot control.

Automation

Robotics

ceojohntran@chatwithrobot.net

+84905311611

© 2025. All rights reserved.

qtran1215@gmail.com