Can a model of a fruit fly’s visual system find its way through a virtual kitchen?
Press Start and watch it happen. The model processes images from the scene and generates neural responses in real time. A controller we built uses those responses to steer the simulated fly around obstacles. Alongside the flight, a glowing network shows how its neural activity changes.
No scripted route. No prerecorded flight. Each step starts with a new image and a fresh steering decision.
We haven’t recreated an entire fruit fly. But we’ve taken a visual system model grounded in biology and connected it to something you can see: a simulated fly navigating a virtual world.
果蠅視覺神經模擬
Amazing GPT6 capability! generate an 3D Blender Mpa from one google map snapshot.
@OpenAI @gptzone_net @sama
Thanks @skirano the hint!
I just reserved my Cloudflare Wallet tag: bluevelo.cloudflare.pay.
Reserve yours now at cloudflare.pay
(東京旅遊)承襲上次的行程規劃，今天手癢PUA一下ChatGPT當當牛馬🦬🐎，看看能不能完成東京訂房🗾
Agoda/Booking/官網訂房，無腦點擊，刷卡收簡訊，下面一位，這種流程，已經人工練習到在手術台上麻醉時，都能用念力完成了(?)
請AI來處理的當然是試圖交叉比對各種訊息混亂化的網站，給他搜出個好價格。初期是手忙腳亂，教也教不會，是要你當牛，不是蝸牛🐌
氣到要去買青草茶來降火
後來才發現GPT當機了🙈，決定原諒他
剛剛把規則說清楚後，發現他比較從從容容，答案能對個80%
等到朽木能夠成木之後，再來決定下一步P
Fishball and Tenpura are two desktop AI companions I built from scratch: local wake gesture → Whisper STT → LLM → streamed TTS with real-time mouth-sync, running on a self-hosted GPU pipeline for near-instant replies. Each one has its own personality baked into the firmware.
Still shipping: graceful power-off, a wake word, and getting them to talk to each other.
#stackchan #m5stack #robot
My StackChan application with LEGO Aquarium set, Can real time answer the local language question with emoji.
#Stackchan #m5stack
