Kashyap Chitta

GitHub - autonomousvision/CaRL: [CoRL 2025] CaRL: Learning Scalable Planning Policies with Simple Rewards [CoRL 2025] CaRL: Learning Scalable Planning Policies with Simple Rewards - autonomousvision/CaRL

Our paper "CaRL: Learning Scalable Planning Policies with Simple Rewards" has been accepted to the Conference on Robot Learning (CoRL 2025).
See you in Seoul at the end of September.

Code & Paper:
github.com/autonomousvi...

At TRI’s newest division: Automated Driving Advanced Development, we’re building a clean-slate, end-to-end autonomy stack. We're hiring, with open roles in learning, infra, and validation: www.tri.global/careers#open...

1/ Can open-data models beat DINOv2? Today we release Franca, a fully open-sourced vision foundation model. Franca with ViT-G backbone matches (and often beats) proprietary models like SigLIPv2, CLIP, DINOv2 on various benchmarks setting a new standard for open-source research.

Adding a nice way to visualize the PPO objective to the rlhf book.
Core for policy-gradient is L is proportional to R*A (R=policy ratio, A = advantage).
PPO makes good actions more likely, up to a point.
PPO makes bad actions less likely, up to a point.

What if the cherry was actually cake? (source: arxiv.org/abs/2506.08007)

🚨Job Alert
W2 (TT W3) Professorship in Computer Science "AI for People & Society"
@saarland-informatics-campus.de/@uni-saarland.de is looking to appoint an outstanding individual in the field of AI for people and society who has made significant contributions in one or more of the following areas:

EurIPS is coming! 📣 Mark your calendar for Dec. 2-7, 2025 in Copenhagen 📅

EurIPS is a community-organized conference where you can present accepted NeurIPS 2025 papers, endorsed by @neuripsconf.bsky.social and @nordicair.bsky.social and is co-developed by @ellis.eu

eurips.cc

In an era of billion-parameter models everywhere, it's incredibly refreshing to see how a fundamental question can be formulated and solved with simple, beautiful math.

- How should we orient a solar panel ☀️🔋? -

Zero AI! If you enjoy math, you'll love this!

Video: www.youtube.com/watch?v=ZKzL...

📢 Present your NeurIPS paper in Europe!

Join EurIPS 2025 + ELLIS UnConference in Copenhagen for in-person talks, posters, workshops and more. Registration opens soon; save the date:

📅 Dec 2–7, 2025
📍 Copenhagen 🇩🇰
🔗eurips.cc

#EurIPS
@euripsconf.bsky.social‬

I really like this paper on relative positional encodings using projective geometry for multi-view transformers, by Li et al. (Berkeley/Nvidia/HKU).

It is elegant: in special situations, it defaults to known baselines like GTA (if identity intrinsics) and RoPE (same cam).

arxiv.org/abs/2507.10496

GitHub - autonomousvision/CaRL: [ArXiv 2025] CaRL: Learning Scalable Planning Policies with Simple Rewards [ArXiv 2025] CaRL: Learning Scalable Planning Policies with Simple Rewards - autonomousvision/CaRL

We have released the code for our work, CaRL: Learning Scalable Planning Policies with Simple Rewards.

The repository contains the first public code base for training RL agents with the CARLA leaderboard 2.0 and nuPlan.

github.com/autonomousvi...

YouTube video by Daniel Dauner CaRL: Learning Scalable Planning Policies with Simple Rewards

In case you find it as relaxing as we do: Here is a 2h+ video of our autonomous RL driving agent CaRL in action! @danieldauner.bsky.social @bernhard-jaeger.bsky.social @kashyap7x.bsky.social
youtube.com/watch?v=_god...

🏟️But we’re not done yet - our workshop continues at #ICCV2025! And the challenge moves forward too, with more prizes and exciting updates. opendrivelab.com/challenge2025/

Do you also plan to open-source this continuous token version? As I understand, the current release is for a discrete-tokenized model set, right?

Our @matthieucord.bsky.social talked on Driving through Generative video Pretraining: VaVIM-VaVAM.
He talked on how we boosted speed & perf. by reducing number of tokens and switching from discrete to continuous tokens

🚗 Excited to announce our RealADSim Workshop & Challenges @ ICCV 2025 — join us!
🏆 Over $40,000 in prizes
🌆 Competition 1: Extrapolated Novel View Synthesis
🚘 Competition 2: Closed-loop Driving in a 3DGS-based Simulator
🎤 Featuring great speakers
🔗 realadsim.github.io/2025/

Consider trying this feed of "Other people also liked"
bsky.app/profile/spac...

If you like ML related posts then you will find what people who liked the same posts also liked in the last 24 hours.

That way you kind of automatically follow people who share your interests.

Robust Autonomy Emerges from Self-Play Self-play has powered breakthroughs in two-player and multi-player games. Here we show that self-play is a surprisingly effective strategy in another domain. We show that robust and naturalistic drivi...

Hiring a postdoc to scale up and deploy RL-based planning onto some self-driving cars! We'll be building on arxiv.org/abs/2502.03349 and learn what the limits and challenges of RL planning are. Shoot me a message if interested and help spread the word please!

Full posting to come in a bit.

YouTube video by Jia-Bin Huang One Step, Big Leap: The Simple Idea Transforming Generative AI

Check out the video to learn this new, elegant formulation of generative models!

youtu.be/swKdn-qT47Q

The 2025 Waymo Challenge results are out! We achieved second place in the End-to-End Driving Challenge with DiffusionLTF and third place in the Scenario Generation Challenge with SHRED. Our approaches prioritized simplicity, and we're preparing to make them publicly available soon.

Congrats to my teammates (@longpollehn.bsky.social, Micha Fauth, @bernhard-jaeger.bsky.social, @danieldauner.bsky.social, @maxigl.bsky.social, @andreasgeiger.bsky.social) and the other top teams!

Reports:
E2E storage.googleapis.com/waymo-upload...
Scenario Gen storage.googleapis.com/waymo-upload...

The 2025 Waymo Challenge results are out! We achieved second place in the End-to-End Driving Challenge with DiffusionLTF and third place in the Scenario Generation Challenge with SHRED. Our approaches prioritized simplicity, and we're preparing to make them publicly available soon.

Foundations of Computer Vision The print version was published by

Our computer vision textbook is now available for free online here:
visionbook.mit.edu

We are working on adding some interactive components like search and (beta) integration with LLMs.

Hope this is useful and feel free to submit Github issues to help us improve the text!

Driving world models become much easier to control and more effective for practical applications when their training data combines human driving demonstrations with synthetic 'non-expert' data from a simulator.

Introducing ReSim: resim-world-model.github.io

🤔 How to reliably simulate future driving scenarios under a wide range of ego behaviors?

😎 Key ingredient: Co-training the world model on heterogeneous data, including real-world data with expert actions and simulated data with non-expert behaviors.

See ReSim: arxiv.org/abs/2506.09981

How can we reliably simulate future driving scenarios under a wide range of ego driving behaviors? Check out: ReSim: Reliable World Simulation for Autonomous Driving
resim-world-model.github.io

Pseudo-Simulation for Autonomous Driving

Pseudo-simulation is a new evaluation paradigm for autonomous vehicles that blends the realism of real-world data with the generalization power of simulation, enabling robust, scalable testing without the need for interactive environments.

WACV'26?

🚗 Pseudo-simulation combines the efficiency of open-loop and robustness of closed-loop evaluation. It uses real data + 3D Gaussian Splatting synthetic views to assess error recovery, achieving strong correlation with closed-loop simulations while requiring 6x less compute. arxiv.org/abs/2506.04218