Ultralow-Power Localization of Insect-Scale Drones: Interplay of Probabilistic Filtering and Compute-in-Memory

Abstract

We propose a novel compute-in-memory (CIM)-based ultralow-power framework for probabilistic localization of insect-scale drones. Localization is a critical subroutine for path planning and rotor control in drones, where a drone is required to continuously estimate its pose (position and orientation) in flying space. The conventional probabilistic localization approaches rely on the 3-D Gaussian mixture model (GMM)-based representation of a 3-D map. A GMM model with hundreds of mixture functions is typically needed to adequately learn and represent the intricacies of the map. Meanwhile, localization using complex GMM map models is computationally intensive. Since insect-scale drones operate under extremely limited area/power budget, continuous localization using GMM models entails much higher operating energy, thereby limiting flying duration and/or size of the drone due to a larger battery.Addressing the computational challenges of localization in an insect-scale drone using a CIM approach, we propose a novel framework of 3-D map representation using a harmonic mean of the "Gaussian-like" mixture (HMGM) model. We show that short-circuit current of a multiinput floating-gate CMOS-based inverter follows the harmonic mean of a Gaussian-like function. Therefore, the likelihood function useful for drone localization can be efficiently implemented by connecting many multiinput inverters in parallel, each programmed with the parameters of the 3-D map model represented as HMGM. When the depth measurements are projected to the input of the implementation, the summed current of the inverters emulates the likelihood of the measurement. We have characterized our approach on an RGB-D scenes dataset. The proposed localization framework is ∼25× energy-efficient than the traditional, 8-bit digital GMM-based processor paving the way for tiny autonomous drones.

Higher order neural processing with input-adaptive dynamic weights on MoS2 memtransistor crossbars

“…Although the crossbar's latency increases, its design and implementation become significantly simplified. Similar memristor and other non-volatile memory-based neural accelerators have also been studied in prior works (Trivedi and Mukhopadhyay 2014;Manasi and Trivedi 2016;Shafiee et al, 2016;Wang et al, 2016;Mikhailenko et al, 2018;Nasrin et al, 2019;Fernando et al, 2020;Ma et al, 2020;Nasrin et al, 2020;Shukla et al, 2021a). However, our subsequent discussion will highlight how dual-gated control of the memtransistor grid can offer unique co-optimization opportunities not available to current memristor-based crossbar designs.…”

ADC/DAC-Free Analog Acceleration of Deep Neural Networks With Frequency Transformation

“…2, zero-padding for channel expansion can be accomplished by injecting a zero input at the padded indices of the input vector. A digital comparator can facilitate soft thresholding, as delineated in (3).…”

“…I N RECENT years, deep learning has gained significant traction in critical domains such as healthcare, finance, security, and autonomous vehicles [1], [2], [3], [4], [5]. Especially as the complexity and accuracy requirements of deep learning applications continue to grow, deploying deep neural networks (DNNs) at the network's edge has become increasingly common for these applications.…”

Insect-inspired AI for autonomous robots

“…These processors are aimed specifically at implementing the parallel, sparse, and asynchronous processing of SNNs (174) and/or exploiting other desirable characteristics of transistors. The latter includes operating in their efficient subthreshold regime (180) or using floating-gate arrays to compute the harmonic mean for low-power localization (181). SNNs have temporal dynamics that more closely model natural neurons.…”