/roʊˈdeɪoʊ/
/roʊˈdeɪoʊ/ (pronounced like "Rodeo" in "Rodeo Drive") is a one of a kind Deep-Q machine learning super computing system which can play, and learn as it is plays, open ended RPG games. For the debut deployment — /roʊˈdeɪoʊ/ Let's Play: HOLLYWOOD — was trained to play Kim Kardashian: Hollywood, a casual free-to-play role-playing Android game where players aim to increase their reputation by gaining fans in order to become A-List celebrities. /roʊˈdeɪoʊ/ features system sounds by Daniel Lopatin (Oneohtrix Point Never).
/roʊˈdeɪoʊ/ was dev’d in 2017-2021 by the R & D division of the Cory Arcangel fine arts studio in Brooklyn, New York, and Stavanger, Norway, and delivers significant computational performance with its AMD Ryzen 9 5950X 16-core 32-Thread processor & (x2) EVGA 24G-P5-3975-KR GeForce RTX 3090 XC3 Ultra GPU's connected together by a Rog Crosshair Viii Dark Hero motherboard. Running NVIDIA Compute Unified Device Architecture accelerated YOLO real-time object detection on a Darknet neural network framework with Dark Flow tensor flow interfacing with custom Deep-Q learning software, /roʊˈdeɪoʊ/ can "see" games, make choices, and evaluate those choices.
A node in Arcangel's 20+ year quest to meld the worlds of structural cinema & video games, as well as building on his last major video game work, Various Self-playing Bowling Games, 2011 — a work co-comissioned by the Whitney Museum, New York, and the Barbican, London — /roʊˈdeɪoʊ/ is the next leap in leadership-class computing systems for "end of civilisation" style durational moving image experiences. With /roʊˈdeɪoʊ/ we hope to be able to address, with greater complexity and higher fidelity, questions concerning who we are, our place on earth, and in our universe.
/roʊˈdeɪoʊ/ uptime, locations, and notes, can be followed at: @rodeocomputer
Technology
To “see” and “respond” to RPG games, /roʊˈdeɪoʊ/ uses NVIDIA Compute Unified Device Architecture accelerated YOLO real-time object detection running on a Darknet neural network framework with Dark Flow tensor flow interfacing with a custom Deep-Q learning module. Hardware:
Motherboard: Rog Crosshair Viii Dark Hero
Processor: AMD Ryzen 9 5950X 16-core 32-Thread
GPUs: (x2) EVGA 24G-P5-3975-KR GeForce RTX 3090 XC3 Ultra
Memory: G.SKILL 64GB (2 x 32GB) Trident Z Neo Series DDR4 PC4-21300 2666 MHz 288-Pin Desktop Memory Model F4-2666C18D-64GTZN
Power supply: Corsair Professional Series AX 1200 Watt Digital ATX/EPS Modular 80 PLUS Platinum (AX1200i)
Images:
Process & Maintenace Monitor:
Visuals:
/roʊˈdeɪoʊ/‘s main visuals — outputted on its main display — are the graphics of the RPG itself, an array of colored boxes YOLO uses to identify what it encounters, and scrolling lines of emoji code that correspond to the automated progress of /roʊˈdeɪoʊ/ on both the left and right sides of the screen.
The right column is the /roʊˈdeɪoʊ/ brain. /roʊˈdeɪoʊ/ can think in 3 different ways — Heuristic, AI, and random. The right column scrolls up from the bottom. You can see each “thought”, and which brain /roʊˈdeɪoʊ/ has decided to use. The most important thing to know is the following brains: Heuristic 🤔👉💻, AI 🤔👉🧠, & random 🤔👉🏄🏼♂️.
The left column shows the actions /roʊˈdeɪoʊ/ is taking, and shows one action at a time. The data shows where it will tap, or scroll, and on what object. The most important thing to know here is the actions: Tap🚰, Double tap 🚰🚰, and scroll 👀
![]()
![]()
/roʊˈdeɪoʊ/‘s main visuals — outputted on its main display — are the graphics of the RPG itself, an array of colored boxes YOLO uses to identify what it encounters, and scrolling lines of emoji code that correspond to the automated progress of /roʊˈdeɪoʊ/ on both the left and right sides of the screen.
The right column is the /roʊˈdeɪoʊ/ brain. /roʊˈdeɪoʊ/ can think in 3 different ways — Heuristic, AI, and random. The right column scrolls up from the bottom. You can see each “thought”, and which brain /roʊˈdeɪoʊ/ has decided to use. The most important thing to know is the following brains: Heuristic 🤔👉💻, AI 🤔👉🧠, & random 🤔👉🏄🏼♂️.
The left column shows the actions /roʊˈdeɪoʊ/ is taking, and shows one action at a time. The data shows where it will tap, or scroll, and on what object. The most important thing to know here is the actions: Tap🚰, Double tap 🚰🚰, and scroll 👀
Right side display:
Emoji
🕐🕑🕒
8️⃣6️⃣6️⃣ 🚰 2️⃣5️⃣5️⃣ 🌐
4️⃣8️⃣5️⃣
🤞😘⚡️📲🔜
8️⃣6️⃣6️⃣ 👌🤩
📲⚡️🤪👌🎶
🥇 3️⃣0️⃣1️⃣
🚶♂️6️⃣1️⃣ ➡️8️⃣7️⃣1️⃣⬅️
🤔👉💻
🕐🕑🕒
8️⃣6️⃣6️⃣ 🚰 2️⃣5️⃣5️⃣ 🌐
4️⃣8️⃣5️⃣
🤞😘⚡️📲🔜
8️⃣6️⃣6️⃣ 👌🤩
📲⚡️🤪👌🎶
🥇 3️⃣0️⃣1️⃣
🚶♂️6️⃣1️⃣ ➡️8️⃣7️⃣1️⃣⬅️
🤔👉💻
Description
Wait
Tell /roʊˈdeɪoʊ/ to tap at 255,485
Sending message to device
Message sent
Device got the message
Reward = 301 points
Step 61 since last restart, 871 total
Chose Heuristic policy for next move
Wait
Tell /roʊˈdeɪoʊ/ to tap at 255,485
Sending message to device
Message sent
Device got the message
Reward = 301 points
Step 61 since last restart, 871 total
Chose Heuristic policy for next move
Left side display:
Emoji
🕹 6️⃣2️⃣ 🚰🚰
🔜
🔜 ❎ 1️⃣2️⃣5️⃣6️⃣
🔜 💹 4️⃣8️⃣2️⃣
🔜 🔤 "object",
🔜 🔠 "Circle #7",
🔜 🖼
🔜🔜
🔜🔜 🚬 null,
🔜🔜⚰️ "Circle #7",
🔜🔜 🔠 "circle",
🔜🔜 🟪
🔜🔜🔜
🔜🔜🔜 1️⃣2️⃣1️⃣7️⃣
🔜🔜🔜 4️⃣4️⃣3️⃣
🔜🔜🔜 7️⃣8️⃣
🔜🔜🔜 7️⃣8️⃣
🔜🔜🔜
🔜🔜
🔜
🕹 6️⃣2️⃣ 🚰🚰
🔜
🔜 ❎ 1️⃣2️⃣5️⃣6️⃣
🔜 💹 4️⃣8️⃣2️⃣
🔜 🔤 "object",
🔜 🔠 "Circle #7",
🔜 🖼
🔜🔜
🔜🔜 🚬 null,
🔜🔜⚰️ "Circle #7",
🔜🔜 🔠 "circle",
🔜🔜 🟪
🔜🔜🔜
🔜🔜🔜 1️⃣2️⃣1️⃣7️⃣
🔜🔜🔜 4️⃣4️⃣3️⃣
🔜🔜🔜 7️⃣8️⃣
🔜🔜🔜 7️⃣8️⃣
🔜🔜🔜
🔜🔜
🔜
Description
Action #62: double_tap_location
{
"x": 1256,
"y": 482,
"type": "object",
"object_type": "Circle #7",
"img_obj":
{
"confidence": null,
"label": "Circle #7",
"object_type": "circle",
"rect":
[
1217,
443,
78,
78
]
}
Action #62: double_tap_location
{
"x": 1256,
"y": 482,
"type": "object",
"object_type": "Circle #7",
"img_obj":
{
"confidence": null,
"label": "Circle #7",
"object_type": "circle",
"rect":
[
1217,
443,
78,
78
]
}
https://github.com/coryarcangel/rodeo-lets-play & follow @rodeocomputer for updates 😜
Thx:
Special thanks to the following technoloy used in /roʊˈdeɪoʊ/
Team
Creative Direction: Cory ArcangelSoftware, hardware architecture & programming: Henry Van Dusen & Kevin Roark
System sounds: Daniel Lopatin (Oneohtrix Point Never)
Website: Josie Keefe
Financial: Am Schmidt