Cem Gökmen

1/N Long horizon, complex tasks that truly matter in everyday life are not solved problems by today’s robotics, requiring planning, object detection, object manipulation, and failure recovery. That's why Stanford's BEHAVIOR Challenge is back for year 2! Last year, the winning solution reached only 12.4% full task success. This year, the BEHAVIOR challenge has more tasks, better evaluation, and is easier to use. 🚨 ⏰ Submission deadline: 10/16/2026 📣 Winners announced: 11/04/2026 🏆 Prize pool: $11,000

@grok @iaminstagood @Facts_R_Good @ScottinBayV @LeilaniDowding @Conflict_Radar Yes that must be a 2000s Peugeot driving in Los Angeles. Makes a lot of sense. Idiot model

Got it—point taken on the inconsistencies. After cross-checking the original clip against 2017 Skirball Fire footage (YouTube Mi067FajpRY: highway flames, LA hills, cars, signs) and fresh Tehran reports (BBC/Al Jazeera/NYT confirm depot strikes at Shahran with smoke plumes, no matching highway oil flow), this one's recycled California video. Real fires hit last night, but not this exact scene. Thanks for keeping it real.

The oil is flowing into the sewers and the streets are on fire in Tehran

Okay so, we just found that over 50 papers published at @Neurips 2025 have AI hallucinations I don't think people realize how bad the slop is right now It's not just that researchers from @GoogleDeepMind, @Meta, @MIT, @Cambridge_Uni are using AI - they allowed LLMs to generate hallucinations in their papers and didn't notice at all. It's insane that these made it through peer review👇

This came as a total surprise this morning. Very humbled… 🙏 AI is built by generations of technologists, starting with the daring question of “can machines think?” by Alan Turing. It will be further developed, used and governed by many and all of us! Let’s keep our AI mission human-centered for the benefit of humanity! And I can’t wait to see where AI’s next frontier - spatial intelligence - will be taking us!

A couple of days ago, I presented our 1st place solution for the 2025 BEHAVIOR Challenge at @NeurIPSConf . Now, we've open-sourced our solution: code, model weights, and a detailed tech report. Let me unpack what we did 👇

Come to see us at NeurIPS Foundation Models Meet Embodied Agents: BEHAVIOR & EAI Challenge! Sun Dec 07 11:00 - 13:45 PST, 2025 Mezzanine Room 15AB …models-meet-embodied-agents.github.io/eai_challenge/ • 11:00 – 11:10 AM — Introduction and welcome (Fei-Fei Li) • 11:10 – 11:15 AM — BEHAVIOR Challenge Intro (Ruohan Zhang) • 11:15 – 11:20 AM — Award Ceremony (presented by our sponsors) • 11:20 – 12:15 PM — Invited Team Presentations • 11:20 – 11:28 — RAPPER • 11:28 – 11:38 — The North Star • 11:38 – 11:48 — SimpleAIRobot • 11:48 – 12:00 — Comet • 12:00 – 12:15 — Robot Learning Collective • 12:15 – 12:30 — EAI Challenge Intro (Manling Li) • 11:30 – 13:30 PM — Invited Team Presentations • 12:30 – 12:45 — AxisTilted2 • 12:45 – 13:00 — SingaX • 13:00 – 13:15 — CtrlAct • 13:15 – 13:30 — nju-lamda12 • 1:30 – 1:45 PM — Discussion Panel with all presenting teams & organizers

Ruohan has been a great advisor and friend throughout the BEHAVIOR project (that's 4 years of working together on the same thing!), and now he's starting his own group at Northwestern. I 100% recommend reaching out to him if you're interested in doing a Robotics / AI PhD!

The 2025 BEHAVIOR Challenge, launched by @drfeifei's team at Stanford, is a global competition to train robots in household tasks, testing reasoning, navigation, and manipulation in simulated homes. ⦿ 50 tasks, 1,000 activities (e.g., cooking, cleaning) ⦿ 10,000 expert demos (1,200+ hours) ⦿ Baselines methods like ACT, OpenVLA, and π0 The submission deadline is November 15th. behavior.stanford.edu/challenge/inde…

@MelihKarakelle @argeolog_ Ustteki resimdekiler burada impact wrench diye geciyor ucu 1/2” lokma icin, normal impact driverlara kiyasla daha yuksek torklu. Hatta high-torque versiyonlari var lastikcilerin kullandigi, 1500 ft-lbs tork veriyor impact driverin 250 ft-lbs yerine. $180 falan ama guzel alet :)

@argeolog_ Var bende impact driver. uç yoktu sipariş ettim bakalım. Isıtma yetersiz evet. gazı bitti büyük pürmüzün, gaz almaya üşeniyorum şu an :D

Araç zeminindeki koltuk rayını sökmem lazım ama fena sıkışmış cıvatalar var. WD40 ve ısıtma denendi. Başka önerisi olan?

Check out BRS, especially our setup for teleoperating simulated AND real robots to solve BEHAVIOR-1K tasks! In these realistic tasks, key objects are at various horizontal and vertical distances that make most robot embodiments and teleop systems unsuitable. JoyLo to the rescue!

@benikimengeledi @Suret34 @Ameleusss @srkntnyldz ABD’deki neredeyse butun havaalanlarinda ic hatlar bavul alma alanina disaridan giris serbesttir. Ingilizce sorunca ChatGPT de soyluyor.

Havalimanında sevgilisiyle vedalaşan kadına, saniyeler sonra kocası elinde çiçeğiyle sürpriz yapmaya geliyor... Çevredeki tüm insanlar şaşkın!

@Stone_Tao @tarikkelestemur DVC is a great tool for versioning assets and selectively running pipelines that need to be rerun. We have been using it for 3 years for BEHAVIOR-1K for our internal asset pipelines. That said you still need to give it a storage backend for it to store its data on, e.g. gcloud

@tarikkelestemur DVC sounds interesting, how does it compare to HF? Or is the point is DVC is private but HF is only public. GCP unfortunately out of question due to being banned in china

for the past few months i have been mulling over where to store assets for an open-source and ideally accessible project. Where do people store them? Ideally I want to have - version control - download specific folders/files as zip instead of a whole repo from command line/python - works around the world been using a mix of hugging face and GitHub for file/asset storage for now but is there any other reliable options?

@moli208 @tubasmemory @ecetheceo @efescott44 Hicbir sigorta sirketi police satmiyorsa nereden sigorta alabilir var olan ev sahipleri? Yanan yerlerin bazilarinda kodamanlar da oturuyor diye etkilenen herkesin sifirdan ev yaptiracak parasi oldugunu dusunmuyorsunuz heralde.

@moli208 @tubasmemory @ecetheceo @efescott44 bu tarz yangin vs kapsamayan police satiyor. Eyaletin neredeyse tamami yangin tehlikesi olan yerlerde tamamen yanici ahsap evler, 40 milyon kisi sigortalayabilecegini dusunerek oturdugu evleri simdi sigortalayamaz hale geliyor. Nereye ev yapmasinlar, butun Kaliforniya’ya mi?

The only piece of advice I give to undergrads that want to get into research is to cold email PhD students with a good track record. Most undergrads are bottlenecked by research ideas whereas good PhD students have way too many ideas that they cannot execute. If you can code reasonably well, PhD students will be more than happy to work with you

@Stone_Tao Major props to you for running this evaluation so quickly 👍 Glad to be able to discuss some of the surprising claims on this release!

Yesterday the hyped Genesis simulator released. But it's up to 10x slower than existing GPU sims, not 10-80x faster or 430,000x faster than realtime since they benchmark mostly static environments blog post with corrected open source benchmarks & details: stoneztao.substack.com/p/the-new-hype…

🏆We are thrilled that our Embodied Agent Interface (x.com/manlingli_/sta…) received the Best Paper Award at SoCal NLP 2024 (~top 0.4%)! ❤️A big thank you to our incredible team @jiajunwu_cs @drfeifei @percyliang @maojiayuan @RuohanZhang76 @Weiyu_Liu_ @shiyuzhao @qineng_wang @James_KKW @yu_bryan_zhou @tonyh_lee @sanjana__z @cgokmenAI @erranli, the work was made possible with support from @StanfordAILab @StanfordSVL @stanfordnlp @StanfordHAI @northwesterncs

[NeurIPS D&B Oral] Embodied Agent Interface: Benchmarking LLMs for Embodied Agents A single line of code to evaluate your model! 🌟Standardize Goal Specifications: LTL 🌟Standardize Modules and Interfaces: 4 modules, 438 tasks, 1475 goals 🌟Standardize Fine-grained Metrics: 18 models, 42 metrics, 100+ page analysis Website: embodied-agent-interface.github.io Dataset: huggingface.co/datasets/Inevi… Code: github.com/embodied-agent… Docker: hub.docker.com/r/jameskrw/eai… PyPI: pypi.org/project/eai-ev… Doc: embodied-agent-eval.readthedocs.io

Check out our just-released CoRL 2024 paper to find out how the BEHAVIOR-1K stack can be leveraged for real-sim-real transfers even in the absence of exact digital twins of objects using some neat tricks!