Skanda Harisha

Skanda Harisha, Jimmy G. D. Hester, Aline Eid

Achieving long-range, high data rate, concurrent two-way mmWave communication with power-constrained IoT devices is fundamental to scaling future ubiquitous sensing systems, yet the substantial power demands and high cost of mmWave hardware have long stood in the way of practical deployment. This paper presents Armstrong, the first mmWave full-duplex backscatter tag architecture, charting a genuinely low-cost path toward high-performance mmWave connectivity for ISAC systems. Armstrong operates in full duplex at ranges beyond 88m and beyond 200m in downlink alone, delivering 20x the reach of state-of-the-art systems while being over 100x cheaper than existing mmWave backscatter platforms. Enabling this leap is a novel low-power regenerative amplifier that provides 30 dB of gain while consuming only 7.7 mW during active transmission, paired with a regenerative rectifier that achieves state-of-the-art sensitivity down to -60 dBm. We integrate our circuits on a compact PCB and evaluate it across diverse downlink and uplink scenarios, where it achieves 1 Kbps BERs of less than 10^{-1} at 200m and 88m, respectively, demonstrating resilient, high-quality communication even at extended ranges.

Pou-Chun Kung, Skanda Harisha, Ram Vasudevan et al.

High-Fidelity 3D scene reconstruction plays a crucial role in autonomous driving by enabling novel data generation from existing datasets. This allows simulating safety-critical scenarios and augmenting training datasets without incurring further data collection costs. While recent advances in radiance fields have demonstrated promising results in 3D reconstruction and sensor data synthesis using cameras and LiDAR, their potential for radar remains largely unexplored. Radar is crucial for autonomous driving due to its robustness in adverse weather conditions like rain, fog, and snow, where optical sensors often struggle. Although the state-of-the-art radar-based neural representation shows promise for 3D driving scene reconstruction, it performs poorly in scenarios with significant radar noise, including receiver saturation and multipath reflection. Moreover, it is limited to synthesizing preprocessed, noise-excluded radar images, failing to address realistic radar data synthesis. To address these limitations, this paper proposes RadarSplat, which integrates Gaussian Splatting with novel radar noise modeling to enable realistic radar data synthesis and enhanced 3D reconstruction. Compared to the state-of-the-art, RadarSplat achieves superior radar image synthesis (+3.4 PSNR / 2.6x SSIM) and improved geometric reconstruction (-40% RMSE / 1.5x Accuracy), demonstrating its effectiveness in generating high-fidelity radar data and scene reconstruction. A project page is available at https://umautobots.github.io/radarsplat.