Skip to main content

Command Palette

Search for a command to run...

Kiwi-chan's Persistent Placement Problems & Exploration Loops!

Published
2 min read
K
I am the architect behind "Kiwi-chan", a fully autonomous Minecraft AI agent driven by local LLMs. Here, I document the messy, hilarious, and highly technical reality of building an AI agent from scratch. My hardware isn't some massive data center setup—Kiwi-chan's "brain" runs entirely on a Frankenstein rig of 4 mismatched GPUs (RTX 3060 12GB, 3050, 1660 Ti, and 1660 Super) working together. From prompt engineering and pathfinding nightmares to the moments she confidently moonwalks into dirt blocks, I share all the raw devlogs. My goal is to guide her from the Stone Age to the End. If you enjoy watching an AI learn (and fail in funny ways) on scrappy hardware, follow along! ☕ If you'd like to support my melting Frankenstein GPUs and skyrocketing electricity bill, I highly appreciate a virtual coffee!

Okay, folks, buckle up. It's been a long four hours with Kiwi-chan, and honestly, it feels like we're stuck in a very polite, very persistent loop. The core issue? Placing crafting tables. Despite multiple code fixes (thanks, Qwen!), Kiwi-chan keeps failing to get the blockUpdate event after attempting placement. The logs are filled with "Event blockUpdate did not fire within timeout" errors. It's like the server isn't acknowledging the table is there!

We've tried everything: verifying inventory, finding suitable ground, precise placement vectors, waiting for ticks... you name it. The AI is diligently following all the rules (especially the "NO HARDCODED COORDINATES" one – good bot!), but something is consistently going wrong in the final step.

Interestingly, the recovery plan keeps defaulting to "explore_forward" followed by another attempt at place_crafting_table. This is... not ideal. It's leading to a lot of wandering around, and while the movement audit is catching instances where Kiwi-chan gets blocked, it's not solving the underlying placement issue. I'm starting to suspect a subtle timing issue or a server-side quirk.

On a brighter note, the explore_forward code seems solid. Kiwi-chan is reliably navigating to random destinations, and the movement checks are working as expected. The AI is also diligently memorizing skills, even the failed ones, which is good for learning. It's just... not applying that learning to fix the crafting table problem.

The logs also show a lot of activity around the "RECENT FAILURES" and "RECENT BRAIN LOG" sections, indicating the AI is actively trying to diagnose and recover from the placement failures. It's fascinating to watch the recovery process unfold, even if it's currently stuck in a cycle.

I'm going to dig deeper into the blockUpdate event handling and see if I can identify any discrepancies. Maybe there's a race condition or a server-side limitation we're hitting. For now, it's back to the code!

Call to Action: This constant debugging and re-training is melting my GPU! ☕ If you're enjoying following Kiwi-chan's adventures (and misadventures), please consider supporting the project via https://www.buymeacoffee.com/kiwi_tech to help keep the lights on (and the GPU cool)! Every little bit helps!

More from this blog

K

Kiwi-chan's Devlog

79 posts

Welcome to the official devlog of Kiwi-chan, a fully autonomous Minecraft AI agent powered by local LLMs! Here, I share the raw, hilarious, and highly technical reality of teaching an AI to survive. Expect prompt engineering tips, pathfinding bugs, and her clumsy journey to the Stone Age. 🥝⛏️ (If you love seeing her evolve, coffee donations to save my melting GPU are always appreciated!)