Computer science (CS) and engineering research both have large and well documented gender diversity gaps. In fact, previous studies have reported that the overall CS Female Author Ratio (FAR) is only in the range of 16-26% and varies significantly between CS subfields ranging from as high as 42% in CS Education to as low as 8% in Theory and Algorithms. To understand the current state of gender diversity of robotics, we recently collected and analyzed the gender of paper authors from all IEEE RAS fully sponsored conferences (ARSO, CASE, HAPTICS, Humanoids, ICRA, ISAM, MEMS, RoboSoft, SIMPAR, SSRR) as well as IROS and RA-L from 2019-2021. Overall, we find that robotics has a long way to go to reach gender parity, with an overall FAR of only 11-12%. We hope this analysis helps the robotics community to continue to emphasize the importance of working to improve our diversity.
In this work, we present the Magnificent Seven Challenges in domain-specific accelerator design that can guide adventurous architects to contribute meaningfully to novel application domains. Although these challenges appear across domains ranging from ML to genomics, we examine them through the lens of autonomous systems as a motivating example in this work. To that end, we identify opportunities for the path forward in a successful domain-specific accelerator design from these challenges.
Model-predictive control (MPC) is a powerful tool for controlling highly dynamic robotic systems subject to complex constraints. However, MPC is computationally demanding, and is often impractical to implement on small, resource-constrained robotic platforms. We present TinyMPC, a high-speed MPC solver with a low memory footprint targeting the microcontrollers common on small robots. Our approach is based on the alternating direction method of multipliers (ADMM) and leverages the structure of the MPC problem for efficiency. We demonstrate TinyMPC both by benchmarking against the state-of-the-art solver OSQP, achieving nearly an order of magnitude speed increase, as well as through hardware experiments on a 27 g quadrotor, demonstrating high-speed trajectory tracking and dynamic obstacle avoidance.
We introduce MPCGPU, a GPU-accelerated, real-time NMPC solver that leverages an accelerated preconditioned conjugate gradient (PCG) linear system solver at its core. We show that MPCGPU increases the scalability and real-time performance of NMPC, solving larger problems, at faster rates. In particular, for tracking tasks using the Kuka IIWA manipulator, MPCGPU is able to scale to kilohertz control rates with trajectories as long as 512 knot points. This is driven by a custom PCG solver which outperforms state-of-the-art, CPU-based, linear system solvers by at least 10x for a majority of solves and 3.6x on average.
In this work we present a new parallel-friendly symmetric stair preconditioner. We prove that our preconditioner has advantageous theoretical properties when used in conjunction with iterative methods for trajectory optimization such as a more clustered eigenvalue spectrum. Numerical experiments with typical trajectory optimization problems reveal that as compared to the best alternative parallel preconditioner from the literature, our symmetric stair preconditioner provides up to a 34% reduction in condition number and up to a 25% reduction in the number of resulting linear system solver iterations.
Perceptive deep reinforcement learning (DRL) has lead to many recent breakthroughs for complex AI systems leveraging image-based input data. However, training these perceptive DRL-enabled systems remains incredibly memory intensive. In this paper, we begin to address this issue through differentially encoded observation spaces. By reinterpreting stored image-based observations as a video, we leverage lossless differential video encoding schemes to compress the replay buffer without impacting training performance. We evaluate our approach with three state-of-the-art DRL algorithms and find that differential image encoding reduces the memory footprint by as much as 14.2x and 16.7x across tasks from the Atari 2600 benchmark and the DeepMind Control Suite (DMC) respectively. These savings also enable large-scale perceptive DRL that previously required paging between flash and RAM to be run entirely in RAM, improving the latency of DMC tasks by as much as 32%..
We introduce RobotPerf, a vendor-agnostic benchmarking suite designed to evaluate robotics computing performance across a diverse range of hardware platforms using ROS 2 as its common baseline. The suite encompasses ROS 2 packages covering the full robotics pipeline and integrates two distinct benchmarking approaches: black-box testing, which measures performance by eliminating upper layers and replacing them with a test application, and grey-box testing, an application-specific measure that observes internal system states with minimal interference. Our benchmarking framework provides ready-to-use tools and is easily adaptable for the assessment of custom ROS 2 computational graphs. Drawing from the knowledge of leading robot architects and system architecture experts, RobotPerf establishes a standardized approach to robotics benchmarking. As an open-source initiative, RobotPerf remains committed to evolving with community input to advance the future of hardware-accelerated robotics.
Embedded machine learning (ML) on low-power devices, also known as "TinyML," enables intelligent applications on accessible hardware and fosters collaboration across disciplines to solve real-world problems. Its interdisciplinary and practical nature makes embedded ML education appealing, but barriers remain that limit its accessibility, especially in developing countries. Challenges include limited open-source software, courseware, models, and datasets that can be used with globally accessible heterogeneous hardware. Our vision is that with concerted effort and partnerships between industry and academia, we can overcome such challenges and enable embedded ML education to empower developers and researchers worldwide to build locally relevant AI solutions on low-cost hardware, increasing diversity and sustainability in the field. Towards this aim, we document efforts made by the TinyML4D community to scale embedded ML education globally through open-source curricula and introductory workshops co-created by international educators. We conclude with calls to action to further develop modular and inclusive resources and transform embedded ML into a truly global gateway to embedded AI skills development.
We present RoboShape, an accelerator framework that leverages two topology-based computational patterns that scale with robot size: (1) topology traversals, and (2) large topology-based matrices. Using these patterns and building on prior work, we expose opportunities to directly use robot topology to inform architectural mechanisms including task scheduling and allocation, data placement, block matrix operations, and sparse I/O data. For the topologically-diverse iiwa manipulator, HyQ quadruped, and Baxter torso robots, RoboShape accelerators on an FPGA provide a 4.0x to 4.4x speedup in compute latency over CPU and a 8.0x to 15.1x speedup over GPU for the dynamics gradients, a key bottleneck preventing online execution of nonlinear optimal motion control for legged robots. Taking a broader view, for topology-based applications, RoboShape enables analysis of performance and resource utilization tradeoffs that will be critical to managing resources across accelerators in future full robotics domain-specific SoCs.
Deep reinforcement learning (DRL) is one of the most powerful tools for synthesizing complex robotic behaviors. But training DRL models is incredibly compute and memory intensive, requiring large training datasets and replay buffers to achieve performant results. This poses a challenge for the next generation of field robots that will need to learn on the edge to adapt to their environment. In this paper, we begin to address this issue through observation space quantization. We evaluate our approach using four simulated robot locomotion tasks and two state-of-the-art DRL algorithms, the on-policy Proximal Policy Optimization (PPO) and off-policy Soft Actor-Critic (SAC) and find that observation space quantization reduces overall memory costs by as much as 4.2x without impacting learning performance.
Historically, computing instructors and researchers have developed a wide variety of tools to support teaching and educational research, including exam and code testing suites and data collection solutions. However, these tools often find limited adoption beyond their creators. As a result, it is common for many of the same functionalities to be re-implemented by different instructional groups within the Computing Education community. We hypothesise that this is due in part to discoverability, availability, and adaptability challenges. Further, instructors often face institutional barriers to deployment, which can include hesitance of institutions to rely on community developed solutions that often lack a centralised authority and may be community or individually maintained. To this end, our working group explored what solutions are currently available, what instructors needed, and the reasons behind the above-mentioned phenomenon. To do so, we reviewed existing literature and surveyed the community to identify the tools that have been developed by the community; the solutions that are currently available and in use by instructors; what features are needed moving forward for classroom and research use; what support for extensions is needed to support further Computing Education research; and what institutional challenges instructors and researchers are currently facing or have faced in using community software solutions. Finally, the working group identified factors that limited adoption of solutions. This work proposes ways to integrate and improve the availability, discoverability, and dissemination of existing community projects, as well as ways to manage and overcome institutional challenges.
We introduce RobotCore, an architecture to integrate hardware acceleration in the widely-used ROS 2 robotics software framework. This architecture is target-agnostic (supports edge, workstation, data center, or cloud targets) and accelerator-agnostic (supports both FPGAs and GPUs). It builds on top of the common ROS 2 build system and tools and is easily portable across different research and commercial solutions through a new firmware layer. We also leverage the Linux Tracing Toolkit next generation (LTTng) for low-overhead real-time tracing and benchmarking. To demonstrate the acceleration enabled by this architecture, we design an intra-FPGA ROS 2 node communication queue to enable faster data flows, and use it in conjunction with FPGA-accelerated nodes to achieve a 24.42% speedup over a CPU.
Tiny robot learning lies at the intersection of embedded systems, robotics, and ML, compounding the challenges of these domains. This paper gives a brief survey of the tiny robot learning space, elaborates on key challenges, and proposes promising opportunities for future work in ML system design.
We introduce and release GRiD, an open-source, GPU-accelerated library for computing rigid body dynamics with analytical gradients. GRiD was designed to accelerate nonlinear trajectory optimization through optimized code generation, GRiD provides as much as a 7.2x speedup over a state-of-the-art, multi-threaded CPU implementation and maintains as much as a 2.5x speedup when accounting for I/O overhead.
This paper introduces RoboRun, a mobile-robot runtime that dynamically exploits the compute-environment synergy to improve performance and energy. We implement RoboRun in the Robot Operating System (ROS) and evaluate it on autonomous drones. We compare RoboRun against a state-of-the-art static design and show 4.5X and 4X improvements in mission time and energy, respectively, as well as a 36% reduction in CPU utilization.
We introduce robomorphic computing; a methodology to transform robot morphology into a customized hardware accelerator morphology. In this work, we (i) present this design methodology; (ii) use the methodology to generate a parameterized accelerator design for the gradient of rigid body dynamics; (iii) evaluate FPGA and synthesized ASIC implementations; and (iv) describe how the design can be automatically customized for other robot models. Our FPGA accelerator achieves speedups of 8x and 86x over CPU and GPU latency, and maintains an overall speedup of 1.9x to 2.9x deployed in an end-to-end coprocessor system. ASIC synthesis indicates an additional factor of 7.2x.
We apply approximate matrix multiplication to artificial Neural Networks for image classification and to the robotics problem of Distributed Simultaneous Localization and Mapping increasing the speed of both applications by 15-20% while maintaining a 97% classification accuracy for NNs running on the MNIST dataset and keeping the average robot position error under 1 meter (vs 0.32 meters for the exact solution). However, both applications experience variance in their results.
We analyze the benefits and tradeoffs of higher degrees of parallelization using a multiple-shooting variant of DDP implemented on a GPU. We describe our implementation strategy and present results demonstrating its performance compared to an equivalent multi-threaded CPU implementation using several benchmark control tasks. Our results suggest that GPU-based solvers can offer increased per-iteration computation time and faster convergence in some cases, but in general tradeoffs exist between convergence behavior and degree of algorithm-level parallelism. This work was [extended](/publication/parallelddp_icra) and used for MPC on a physical Kuka arm.
We build on recent work on Unscented Dynamic Programming (UDP) - which eliminates dynamics derivative computations in DDP - to support general nonlinear state and input constraints using an augmented Lagrangian. The resulting algorithm has the same computational cost as first-order penalty-based DDP variants, but can achieve constraint satisfaction to high precision without the numerical ill-conditioning associated with penalty methods.
We describe the pedagogy behind the MIT Beaver Works Summer Institute Robotics Program, a new high-school STEM program in robotics. The program utilizes state-of-the-art sensors and embedded computers for mobile robotics. The program was offered as a four-week residential program at MIT in the summer of 2016.