NVIDIA Jetson Orin Nano 8GB
PulseAugur coverage of NVIDIA Jetson Orin Nano 8GB — every cluster mentioning NVIDIA Jetson Orin Nano 8GB across labs, papers, and developer communities, ranked by signal.
12 day(s) with sentiment data
-
Lightweight HMMF-Net achieves high accuracy in fish feeding assessment
Researchers have developed HMMF-Net, a novel lightweight network designed for accurate fish feeding intensity assessment in aquaculture. This multimodal architecture effectively combines visual and acoustic data, overco…
-
Energy-aware OVD scheduler boosts robot autonomy and efficiency
Researchers have developed an energy-aware scheduling system for open-vocabulary object detection (OVD) on low-power mobile robots. This system dynamically adjusts the YOLO-World-S detection modes based on factors like …
-
Banana Pi BPI-SM10 offers 60 TOPS RISC-V alternative to NVIDIA Jetson Orin Nano
The Banana Pi BPI-SM10 is a new RISC-V development board that offers 60 TOPS of performance, positioning it as an alternative to NVIDIA's Jetson Orin Nano. This board is designed for embedded AI applications and aims to…
-
New ORBITAL framework optimizes BCI compression for robustness and efficiency
Researchers have developed ORBITAL, a new framework designed to optimize brain-computer interfaces (BCIs) by compressing electroencephalography (EEG) decoders while maintaining adversarial robustness. This method combin…
-
New Bi-Mamba Model Enhances Robot Navigation with Real-Time Trajectory Prediction
Researchers have developed TrajMamba, a novel bidirectional state space model designed for real-time human trajectory prediction using RGB-D cameras. This model addresses the train-deploy pipeline gap by aligning pose e…
-
NVIDIA releases Nemotron 3.5 Bolt E4B for local AI agents
NVIDIA has released Nemotron 3.5 Bolt E4B 4GB, an open small language model designed for efficient local operation on devices like voice assistants, game characters, and robots. This model is post-trained from Google's …
-
New framework SLoSa trains small, safe AI models for robot task planning
Researchers have developed SLoSa, a framework designed to create small, energy-efficient, and safety-aware foundation models for robotic task planning. This approach combines classical planning methods with large langua…
-
Efficient U-Net model developed for rapid spinal CT segmentation
Researchers have developed SpineContextResUNet, a computationally efficient 3D Residual U-Net designed for rapid spinal CT segmentation. This model addresses the high resource demands of current state-of-the-art methods…
-
Greenpeace protesters disrupt Nvidia GTC keynote over drone chip concerns
During Jensen Huang's Nvidia GTC keynote in Berlin, Greenpeace protesters disrupted the event, demanding answers about Nvidia chips found in Russian drones used in Ukraine and the company's broader environmental impact.…
-
Edge GPU AI performance bottlenecked by memory bandwidth, not FLOPs
A new research paper highlights that the performance of query-based segmentation decoders on edge GPUs is primarily limited by memory bandwidth rather than computational power (FLOPs). The study, which profiles models l…
-
EdgeWise slashes AI cloud costs by 87% with selective LLM delegation
EdgeWise has developed a novel approach to reduce cloud computing costs and latency for AI agents by selectively delegating tasks. Their system, detailed in an arXiv paper, uses a small "step-selector" model running on …
-
FPGA-based FALCON system ensures latency guarantees for autonomous driving
Researchers have developed FALCON, a novel system for camera-based autonomous driving that prioritizes latency-certifiable occupancy networks. This system, implemented on an AMD Kria KR260 FPGA, ensures that latency gua…
-
Edge AI system DefoliationCAM assesses grapevine pest damage in real-time
Researchers have developed DefoliationCAM (D-CAM), an edge AI system designed to assess insect defoliation severity in grapevines in real-time. The system utilizes a YOLOX-Nano model for cordon localization and a lightw…
-
New 3D Road User Estimation Method Uses Camera-Radar Fusion
Researchers have developed a novel two-stage approach for estimating the 3D shape and orientation of road users from roadside RGB cameras and 3D radar. This method addresses the challenges of adapting existing end-to-en…
-
New quantization method boosts Mamba super-resolution speed by 2.23x
Researchers have developed a new post-training quantization (PTQ) framework called Token-Group Quantization for Mamba-based super-resolution models. This method addresses the challenge of varying activation ranges acros…
-
Gemma E4B optimized for Jetson Orin Nano with new EXL3 engine
A new engine called little-gemma V1.0 has been developed for the NVIDIA Jetson Orin Nano, offering improved performance over llama.cpp for the Gemma E4B model. The developer forked exllamav3 and ported code from little-…
-
Liquid AI open-sources d1 decision models for edge AI
Liquid AI has open-sourced two of its d1 decision models, d1-3B and d1-omni-600M, making them available on Hugging Face. These models are designed to process various inputs like text, images, and audio to provide struct…
-
New protocol evaluates pedestrian tracking systems beyond aggregate scores
A new evaluation protocol for pedestrian-centric tracking systems has been developed to assess performance beyond aggregate scores. This protocol isolates tracker behavior by using shared detections and specifically eva…
-
Kinematic MeanFlow enhances robotic model action generation speed
Researchers have developed Kinematic MeanFlow (K-MF), a novel one-step action generation policy designed to improve the efficiency of Robotic Foundation Models (RFMs). Traditional multi-step flow matching methods suffer…
-
New VLA-ULAP system boosts edge AI efficiency by reducing cloud calls
Researchers have developed VLA-ULAP, a system that combines cloud-based vision-language-action (VLA) models with a lightweight local predictor to improve efficiency and speed for edge devices. This approach significantl…