OmniEVA: Bridging the 2D–3D Gap in Embodied AI

New paper introduces OmniEVA, a versatile embodied planner that pushes the boundaries of multimodal large language models (MLLMs) for robotics and spatial reasoning.

Results: OmniEVA achieves state-of-the-art performance across 2D/3D reasoning benchmarks and outperforms existing models in object navigation tasks.

Paper: https://arxiv.org/pdf/2509.09332v1
Project: https://omnieva.github.io/

#EmbodiedAI #Robotics #LLM #MLLM #3DVision #AIResearch #AIPlanning

OpenAI confirms GPT-6 is in fast track development. SMBs must align budgets & upskill teams now for seamless AI adoption. #AIPlanning #SMBTech

https://www.bleepingcomputer.com/news/artificial-intelligence/openai-says-gpt-6-is-coming-and-itll-be-better-than-gpt-5-obviously/

If you need efficient constraint solving with long-term support, Timefold is the future. Less waiting, more optimizing. 🚀 #Optimization #Timefold #AIPlanning

Having used OptaPlanner in multiple projects and now testing Timefold, I can confidently say it outperforms OptaPlanner in every way. The devs are active and receptive while Red Hat seems to have quietly shelved OptaPlanner moving it to the KIE / Apache team who do not offer support.

https://timefold.ai/

RAG-Modulo: Solving Sequential Tasks using Experience, Critics, and Language Models: https://arxiv.org/pdf/2409.12294

We saw how #LLMs and #planners #aiplanning help each to produce behavior. It also works with data: coupling #GenAI with an actual #database may yield better results. This is called #RAG.

This paper shows an application of the state-of-the-art of using RAG for interactive #robotics, by proposing a framework. It is not mature, but it lays down the architecture to get started.

Meet Majid Khadiv, our newly appointed Professor of #AIPlanning in #DynamicEnvironments, in the latest "NewIn" episode. His research focuses on how to enable #robots to perform tasks that are dangerous for humans, such as putting out fires: http://go.tum.de/413334 🤖

📷 A.Heddergott

▶️https://youtu.be/sLdfmLYQJK4
📹@prolehre

I have a #proceduralgeneration #procgen idea for puzzles/quests/general sequence of actions in #videogames that im too burned out on programming currently to implement using the formalisms of #ai #AIPlanning

So a puzzle is a sequence of steps (a plan) in a defined ruleset.so we formalize the ruleset in a planner language,set out some key actions that we want to maybe have in the plan,search for a plan first then generate the world in which the plan is possible. 1/2