William Fu

Srivatsan Krishnan, Behzad Boroujerdian, William Fu et al.

We introduce Air Learning, an open-source simulator, and a gym environment for deep reinforcement learning research on resource-constrained aerial robots. Equipped with domain randomization, Air Learning exposes a UAV agent to a diverse set of challenging scenarios. We seed the toolset with point-to-point obstacle avoidance tasks in three different environments and Deep Q Networks (DQN) and Proximal Policy Optimization (PPO) trainers. Air Learning assesses the policies' performance under various quality-of-flight (QoF) metrics, such as the energy consumed, endurance, and the average trajectory length, on resource-constrained embedded platforms like a Raspberry Pi. We find that the trajectories on an embedded Ras-Pi are vastly different from those predicted on a high-end desktop system, resulting in up to 40% longer trajectories in one of the environments. To understand the source of such discrepancies, we use Air Learning to artificially degrade high-end desktop performance to mimic what happens on a low-end embedded system. We then propose a mitigation technique that uses the hardware-in-the-loop to determine the latency distribution of running the policy on the target platform (onboard compute on the aerial robot). A randomly sampled latency from the latency distribution is then added as an artificial delay within the training loop. Training the policy with artificial delays allows us to minimize the hardware gap (discrepancy in the flight time metric reduced from 37.73% to 0.5%). Thus, Air Learning with hardware-in-the-loop characterizes those differences and exposes how the onboard compute's choice affects the aerial robot's performance. We also conduct reliability studies to assess the effect of sensor failures on the learned policies. All put together, Air Learning enables a broad class of deep RL research on UAVs. The source code is available at:http://bit.ly/2JNAVb6.

Colby R. Banbury, Vijay Janapa Reddi, Max Lam et al.

Recent advancements in ultra-low-power machine learning (TinyML) hardware promises to unlock an entirely new class of smart applications. However, continued progress is limited by the lack of a widely accepted benchmark for these systems. Benchmarking allows us to measure and thereby systematically compare, evaluate, and improve the performance of systems and is therefore fundamental to a field reaching maturity. In this position paper, we present the current landscape of TinyML and discuss the challenges and direction towards developing a fair and useful hardware benchmark for TinyML workloads. Furthermore, we present our four benchmarks and discuss our selection methodology. Our viewpoints reflect the collective thoughts of the TinyMLPerf working group that is comprised of over 30 organizations.

Bardienus P. Duisterhof, Srivatsan Krishnan, Jonathan J. Cruz et al.

We present fully autonomous source seeking onboard a highly constrained nano quadcopter, by contributing application-specific system and observation feature design to enable inference of a deep-RL policy onboard a nano quadcopter. Our deep-RL algorithm finds a high-performance solution to a challenging problem, even in presence of high noise levels and generalizes across real and simulation environments with different obstacle configurations. We verify our approach with simulation and in-field testing on a Bitcraze CrazyFlie using only the cheap and ubiquitous Cortex-M4 microcontroller unit. The results show that by end-to-end application-specific system design, our contribution consumes almost three times less additional power, as compared to competing learning-based navigation approach onboard a nano quadcopter. Thanks to our observation space, which we carefully design within the resource constraints, our solution achieves a 94% success rate in cluttered and randomized test environments, as compared to the previously achieved 80%. We also compare our strategy to a simple finite state machine (FSM), geared towards efficient exploration, and demonstrate that our policy is more robust and resilient at obstacle avoidance as well as up to 70% more efficient in source seeking. To this end, we contribute a cheap and lightweight end-to-end tiny robot learning (tinyRL) solution, running onboard a nano quadcopter, that proves to be robust and efficient in a challenging task using limited sensory input.

William Fu, Raymond Lin, Daniel Inge

Fully homomorphic encryption has allowed devices to outsource computation to third parties while preserving the secrecy of the data being computed on. Many images contain sensitive information and are commonly sent to cloud services to encode images for different devices. We implement image processing homomorphically that ensures secrecy of the image while also providing reasonable overhead. We first present some previous related work, as well as the fully homomorphic encryption scheme we use. Then, we introduce our schemes for JPEG encoding and decoding, as well as schemes for bilinear and bicubic image resizing, as well as some data and analysis of our homomorphic schemes. Finally, we outline several issues with the homomorphic evaluation of proprietary algorithms, and how a third party can gain information on the algorithm through noise.

William Fu, Raymond Lin, Daniel Inge

WebAssembly (wasm) has recently emerged as a promisingly portable, size-efficient, fast, and safe binary format for the web. As WebAssembly can interact freely with JavaScript libraries, this gives rise to a potential for undesirable behavior to occur. It is therefore important to be able to detect when this might happen. A way to do this is through taint tracking, where we follow the flow of information by applying taint labels to data. In this paper, we describe TaintAssembly, a taint tracking engine for interpreted WebAssembly, that we have created by modifying the V8 JavaScript engine. We implement basic taint tracking functionality, taint in linear memory, and a probabilistic variant of taint. We then benchmark our TaintAssembly engine by incorporating it into a Chromium build and running it on custom test scripts and various real world WebAssembly applications. We find that our modifications to the V8 engine do not incur significant overhead with respect to vanilla V8's interpreted WebAssembly, making TaintAssembly suitable for development and debugging.