Elon Musks Optimus 3.0 Tesla Robot Learning in Tesla Diner Gemini 3.0 LEAK Amazon Kiro AI IDE

Nevertheless, these diners may serve as a sort of extra training environment for Tesla Optimus humanoids to help them develop and refine home robot skills to be included in commercial releases later on. There are also reports from Chinese suppliers of Tesla parts that the robot is undergoing a redesign, with these manufacturers expecting orders for robot parts to resume in around two months. And next to the humanoids, the Tesla Diner features multiple 45-foot movie screens outdoors, allowing patrons to watch films while charging their electric vehicles. As for power, the main supercharger area accommodates around 35 vehicles at once, with everything from burgers to sandwiches to even fried chicken and more being offered on the food menu.

Meanwhile, Engineered Arts just announced a significant new feature for its robot called Rolls, providing an intuitive way to define the robot’s behavior, preferred interaction style and knowledge base according to different situations. To get started, users simply access their robot’s web-based user interface and log into their Tritium account, where the Rolls interface is now available. There are currently five core tabs within Rolls, identity for basic details, languages to set preferred speech and voices, personality to adjust interaction style or choose from presets, knowledge for adding specialist company or product information, and abilities where users can toggle specific capabilities as needed.

On top of this, Rolls allows for quick adjustments, where users can change the robot’s name, home location, or limit spoken languages to just one or multiple options. Furthermore, it is possible to assign a personalized set of skills, such as enhancing vision capability or the ability to recognize voices for an improved customer experience, all without any coding required. By simplifying customization and expanding flexibility, the company hopes to drive wider adoption of humanoid robots in customer service, retail, and specialized tasks, where personal interaction and expertise are key. But robots are also now capable of growing themselves by simply recycling parts from others.

This was just demonstrated at Columbia University, with researchers having unveiled this breakthrough process that lets robots physically grow by absorbing materials from their environment or even reusing parts from fellow machines. Detailed in their study published in Science Advances, the new technique, which they call robot metabolism, takes inspiration from biology, enabling robots to self-assemble, repair, and adapt in response to new needs. Additionally, the team has demonstrated what’s called the truss link, a modular, magnetic robot based on the simple bar-shaped connectors, but with a twist. That’s because each truss link can freely connect to others at various angles, forming complex shapes and three-dimensional robots that can morph and adapt.

On top of this, these robots are able to physically integrate new parts. For example, a robot shaped like a tetrahedron was able to add another link and use it like a walking stick, speeding up its descent by more than 66.5%. Furthermore, roboticists believe that creating machines capable of such self-directed growth is the beginning of resilient, self-sustaining roboticologies, setting a clear path for evolving autonomous machines in the future. And when it comes to upcoming AI tech, Gemini 3.0 is shaping up to be the next major milestone for Google DeepMind’s large language models. In fact, references to internal versions, such as Beta 3 Pro and Beta 3 Flash, have now surfaced in the open-source Gemini command line tool, indicating final preparations for a feature-rich upgrade.

Notably, Gemini 3.0 introduces an advanced DeepThink reasoning architecture. This system aims to take on complex, multi-step reasoning tasks more accurately, plus it comes with built-in verifier reasoning by default, streamlining output quality, and reducing user intervention. On top of this, Gemini 3.0 will extend its multimodal input capabilities. Whereas Gemini 2.5 already processed text, images, audio, and short videos, this new release is expected to support real-time video at up to 60 frames per second, plus 3D object understanding and geospatial analysis. Furthermore, its context window will reportedly expand into the multi-million token range, allowing the model to handle extremely long documents while maintaining coherent memory and retrieval.

In terms of performance, Gemini 3.0 will leverage Google’s next-generation TPUV 5P accelerators to ensure nearly real-time responses. The current timeline, based on Google’s annual release cadence, points to a likely December 2025 rollout. And finally, with the launch of Kero, Amazon is introducing a new contender in the fast-evolving landscape of artificial intelligence software tools. The company has rolled out Kero, a fully AI-powered integrated development environment in public preview. But what sets Kero apart is its ambition to tackle the growing problem of under-documented, AI-generated software, an increasingly common headache as companies accelerate their development pipelines with artificial intelligence.

Unlike traditional coding assistants, Kero deploys autonomous AI agents designed to not only help write code, but also automatically generate, update and track technical documentation and project plans as work progresses. On top of this, it aims to create seamless transitions from rapidly built prototypes to production-level systems with robust requirements, proper testing protocols, and centralized up-to-date documentation. Furthermore, Kero stands apart from peers, such as GitHub’s Agent Mode and Google’s Gemini Code Assist, by breaking down a developer’s intent into structured, trackable components. Each prompt is deconstructed into requirements, design docs, and task lists, which are dynamically updated as code changes.

Plus, automated checks are run whenever files are saved or modified, with the system handling critical tasks like updating documentation and scanning for inconsistencies. Though Kero is developed by an Amazon Web Services team, its launch avoids prominent Amazon branding, with just subtle references such as an AWS logo in the footer. For now, Kero is available without charge during its preview phase, with paid tiers planned, a free tier capped at 50 agent interactions per month, a pro offering at $19 monthly for up to 1,000 interactions, and a pro plus tier costing $39 per month for 3,000 interactions.

Anyways, make sure to like and subscribe and tell us whether or not the CL3 is a good dancer. [tr:trw].