DARPA Grand Challenge Research Articles

Understanding the long-term emergence of autonomous vehicles technologies

Identifying emerging technologies has been of long-standing interest to many scholars and practitioners. Previous studies have introduced methods to capture the concept of emergence from bibliographic records, including the recently proposed Technology Emergence Indicator (Carley et al. 2018). This indicator method has shown to be applicable to various technological fields. However, the indicator uses a limited time window, which can overlook the potential long-term evolution of emerging technologies. Moreover, the existing method suffers from interpretability, because it can be difficult to understand the context in which identified emerging terms are used. In this paper, we propose an improved version of the Technology Emergence Indicator that addresses these issues. In doing so, we examine emerging topics within the field of autonomous vehicles technologies during the period of 1991-2018, guided by a proposition about the long-term diffusion of an emerging technology topic. The results show that different autonomous vehicle technology topics emerge during each of the three 10-year periods under analysis, including an initial period of understanding the surrounding environment and path planning, a second period marked by DARPA Grand Challenge motivated factors associated with the urban environment and communication technologies, and a third period relating to machine learning and object detection. This association with certain emerging technology topics in each decade is also characterized by different trajectories of continued or cyclical carryover across the decades. The results suggest a methodology that practitioners can use in examining research areas to understand which topics are likely to persist into the future.

Blaming Automated Vehicles in Difficult Situations

The third driverless car competition of the DARPA Grand Challenge (Urban Challenge) in 2007 saw six autonomous vehicle teams finishing the event successfully. Since then, Automated Vehicles (AVs) made huge strides towards deployment on a large scale. Despite all this progress, AVs continue to make mistakes, some of which have resulted in the deaths of passengers and pedestrians. These crashes received wide coverage in the media and drew a parallel bleak picture on the public’s lack of enthusiasm for this technology. However, not all mistakes are equal. While some mistakes are avoidable, others are hard to avoid even by highly-experienced professional drivers. As they continue to shape citizens’ attitudes towards AVs, we need to understand whether people differentiate between different types of error, and whether these are treated proportionally. In this paper, we ask the following two questions: 1) when an automated car makes a mistake, does the perceived difficulty or novelty of the situation predict blame attributed to it? How does that blame attribution compare to a human driving a regular car? Through two studies we find that the amount of blame people attribute to machine drivers and human drivers is sensitive to the difficulty of the situation. However, while some situations could be more difficult for machine drivers and others are harder for human drivers, people blamed machine drivers more, regardless. Our results provide insights on a crucial, yet under-studied, angle in understanding psychological barriers impeding the public’s adoption of AVs.

AFLPro: Direction sensitive fuzzing

Fuzzing is a simple and popular technique that has been widely used to detect vulnerabilities in software. However, due to its blind mutation, fuzzing brings many limitations. First, it is difficult for fuzzing to pass the sanity checks, which makes fuzzing unable to target vulnerability or crash locations effectively. Secondly, blind mutation limits the diversity of seed generation and makes it difficult for the fuzzing process to achieve convergence.In this paper, we propose a direction sensitive fuzzing solution AFLPro. On the one hand, it focuses on seed selection, using a new fuzzing scheme based on Basic Block Aggregation (BBA), which reduces the possibility of seed selection in the wrong direction. By applying a multi-dimensional oriented seed selection strategy, it achieves fine-grained seed selection. On the other hand, based on biological evolution, AFLPro optimizes genetic variation to ensure the diversity of seed varieties and the convergence of fuzzing tests. Besides, AFLPro also incorporates lightweight static analysis to obtain information about the target program (this paper only studies closed source programs), providing complete semantic guidance for fuzzing through resource integration.We implemented a prototype of AFLPro based on the popular fuzzer AFL. We evaluated it on three datasets: DARPA Grand Challenges (CGC), LAVA-M dataset, and a set of real-world applications. The results show that in 92% of all three datasets, AFLPro exhibits better vulnerability detection capabilities than all of the state-of-the-art fuzzers mentioned in this paper.

Al-enabled spectrum technology: What's next? - [Spectral Lines

You've graduated from the school of spectral hard knocks, Paul Tilghman, a U.S. Defense Advanced Research Projects Agency program manager, told the teams competing in DARPA's Spectrum Collaboration Challenge (SC2) finale on 23 October. The three-year competition had just concluded. ¶ Hard knocks wasn't an exaggeration—the 10 teams that made it to the finale, as well as others who were eliminated in earlier rounds of the competition, had been tasked with doing something that had never been done before. Their challenge was to see if AI-managed radio systems could work together to share wireless spectrum more effectively than static, preallocated bands. They had spent three years battling it out in matchups in Colosseum, an RF emulator that DARPA built specially for the competition.¶ By the end, the top teams had demonstrated that their systems could transmit more data over less spectrum than is possible using existing standards like LTE; they also showed an impressive ability to reuse spectrum over multiple radios. In some matchups, the radio systems of five teams were transmitting 200 or 300 percent more data than is possible with today's rigid spectrum band allocations. And that's important, given that we're facing a looming wireless-spectrum crunch. _ But, as Tilghman also stressed during SC2, when a DARPA Grand Challenge ends, it doesn't mean the technology is ready to go to market. These challenges are more about proving that a new technological idea is possible. So the contest showed that AI-controlled radios can work together to share spectrum among themselves, and pack more data into a given amount of spectrum. But where does the technology go from here, now that the DARPA challenge is finished?

Will rule the airwaves: A DARPA grand challenge seeks autonomous radios to manage the wireless spectrum

IN THE EARLY 2000S, BLUETOOTH ALMOST MET AN UNTIMELY END. The first Bluetooth devices struggled to avoid interfering with Wi-Fi routers, a higher-powered, more-established cohort on the radio spectrum, with which Bluetooth devices shared frequencies. Bluetooth engineers eventually modified their standard—and saved their wireless tech from early extinction—by developing frequency-hopping techniques for Bluetooth devices, which shifted operation to unoccupied bands upon detecting Wi-Fi signals. • Frequency hopping is just one way to avoid interference, a problem that has plagued radio since its beginning. Long ago, regulators learned to manage spectrum so that in the emerging wireless ecosystem, different radio users were allocated different frequencies for their exclusive use. While this practice avoids the challenges of detecting transmissions and shifting frequencies on the fly, it makes very inefficient use of spectrum, as portions lay fallow. • Today, demand is soaring for the finite resource of radio spectrum. Over the last several years, wireless data transmission has grown by roughly 50 percent per year, driven largely by people streaming videos and scrolling through social media on their smart-phones. To meet this demand, we must allocate spectrum as efficiently as possible. Increasingly, that means that wireless technologies cannot have exclusive frequencies, but rather must share available spectrum. Frequency hopping, which Bluetooth uses, will be part of the solution, but to cope with the surging demand we are going to have to go far beyond it.

Special Issue on Real World Robot Challenge in Tsukuba – Technology for the Coexistence with Human Beings –

Robotics researchers appear to have shifted their focus since the Great East Japan Earthquake. Despite a large-scale national R&D project on Advanced Robots for Hazardous Environments for 8 years from 1983, the technology has not been put to use effectively following the Great Hanshin Earthquake and researchers are starting to look more closely at applications enabling robots to replace human beings, e.g., by rescuing victims of earthquake and flood disasters. A good example is Quince and the remote-controlled heavy equipment used to handle the pyroclastic flow at Unzen volcano in 1991 and then mobilized at the Fukushima Daiichi nuclear power plant following the Great East Japan Earthquake and tsunami in 2011. Robotics researchers of the past only used to describe their dream proposals when asked what robots could accomplish. Since the 2011 disaster event, however, it has become clear that the issues we should be looking at are those of developing robots that are practical and useful. If we are asked what role the real-world competition Tsukuba Challenge plays in this context, would first recommend that those who hope to take part visit the actual site and see for themselves what they must do to field a winning robot. The Tsukuba Challenge site includes public thoroughfares traversed by pedestrians and cyclists. Although there are no cars or motorcycles on these paths, almost anything can happen. From hot summer until cooler fall when the official run is held daily temperature, rain, wind and typhoons, and trees – all of which must be detected for navigation – undergo many changes. These changes require that robots navigate accurately in this real-world environment. In general, robotics research papers are accepted for publication if robots can navigate as planned in a restricted and fixed mobile environment even if they can do so only once. In the Tsukuba Challenge, however, many things can go wrong and robot maneuvers can become unstable or even nonmobile, regardless of successful navigation in laboratory settings. There is no space here to discuss all possible factors, but the established navigation method by one paper is not always successful and responsible in the actual Tsukuba Challenge environment. Robots historically came to be as devices operated by human beings, but those taking part in the Challenge must be controlled by computers instead of human operators. This means that it must be confirmed that robots can operate as required in experiments meeting various conditions. Teams may conduct ten or so trial runs a year on the actual route so that their systems can be adequately adjusted and modified. The total process that competing in the Tsukuba Challenge requires also has an educational effect in grooming new talent in robotics. The Challenge differs from the DARPA Grand Challenge and Urban Challenge, held from 2004 to 2007, in the size – robots must not be too big and must not appear threatening – and the need to take the presence of human beings into consideration. Although the task from 2007 to 2012 had been only to navigate a preset route and reach a final goal, still the percentage of successful runs has declined. From 2013 on, another task has been added – that of locating specific persons (search targets) within given areas – to encourage the use of advanced technology in realizing useful robots. Those interested may see the results on the Tsukuba Challenge website as follow: http://www.tsukubachallenge.jp/tc2013. The selection of articles for this special issue emphasized the following criteria: 1) For robots successfully completing required tasks, describing and discussing the superiority of the control technology and results. 2) For unsuccessful robots, clearly analyzing how the actual run differed from researchers’ expectations and pinpointing the underlying causes of failure. The submitted papers describe technologies that have enabled robots to navigate in spaces shared with human beings, and we hope that a study of these papers will spur readers to accelerate advances in autonomous mobile robots.

The Next Generation of Unmanned Ground Vehicles

Over the last decade, the development of Unmanned Ground Vehicles (UGVs) has received significant attention through technology competitions, such as the DARPA Grand Challenges, where an unmanned vehicle autonomously navigated across a desert or through semiurban roads. Although this marked a significant step forward in autonomous navigation, the current generation of UGVs are only capable of operating in controlled environments where the dynamics of a scene are well-understood. For example in both of the DARPA Grand Challenges, the participants were given detailed maps and the environment was carefully controlled during the competition. The next generation of UGVs will need to operate in uncontrolled environments where the terrain and infrastructure is uncertain and humans could be present. This paper discusses the challenges and current developments in the areas of sensing, localisation and planning to realise the next generation of UGVs.

5.2.2 Normative and Descriptive Models for Test & Evaluation of Unmanned and Autonomous Systems of Systems

AbstractThe United States Department of Defense (DoD) has purchased and deployed many unmanned autonomous systems and is expected to do so at an ever‐increasing rate in the years to come. These autonomous systems are often expected to operate in system of systems environments. Our research aims to address a number of questions posed by members of DoD's acquisition workforce regarding test and evaluation of these systems.In this paper we present a small set of normative (ideal) and descriptive (actual) models for test and evaluation. We describe test strategies used in the DARPA Grand Challenge competition and the SPHERES test bed, and extract lessons learned from each case study. We will use these models as a step toward identifying prescriptive models for testing unmanned and autonomous systems of systems. The prescriptive models will be implemented within PATFrame (Prescriptive and Adaptive Testing Framework), the decision support system that we are developing.

SIGEST

In recent years, almost everyone has heard about autonomous vehicles. The well-known DARPA grand challenges of the last two years concerned autonomous vehicles navigating challenging courses and terrain. The intent of developing this technology is that one day vehicles will be able to operate autonomously in a variety of dangerous situations, from military scenarios to disaster relief. How will these vehicles communicate and deploy to cover a space evenly, in an environment where they may only have local knowledge of other vehicles? This is an example of the problem that motivates this issue's SIGEST paper. The paper “Nonsmooth Coordination and Geometric Optimization via Distributed Dynamical Systems,” by Jorge Cortés and Francesco Bullo, first appeared in the SIAM Journal on Control and Optimization, 44 (2005), pp. 1543–1574. In this seminal work, the authors design distributed (i.e., reliant on local knowledge) coordination algorithms for dynamic networks, and provide formal verification of the correctness of these algorithms. One of the important and challenging aspects of the problem they study is that they assume that the communication topology of the network may change as the system evolves, as opposed to remaining fixed. The paper considers two different scenarios for “even spacing” in the dynamic network, which are easy to describe informally if we continue to consider the example of vehicles in a convex polygonal region. The first is to cause each point in the region to be as close as possible to the nearest vehicle (akin in a static sense to positioning post offices in a town to minimize the longest distance of any home to the nearest post office); the second is to require each vehicle to be as distant as possible from all other vehicles, and the boundary of the region. In the case of just one vehicle, the first scenario corresponds to finding the center of the smallest circle entirely containing the polygon, whereas the second corresponds to finding the center of the largest circle fully contained within the polygon. It turns out that these disk-covering and sphere-packing concepts are related to the solutions in general. One of the fascinating and beautiful aspects of this paper is that it closely combines three rather distinct areas of applied mathematics: geometry and in particular Voronoi diagrams; dynamical systems where relationships can evolve and change; and optimization, specifically nonsmooth optimization, meaning that the objective functions are not continuously differentiable. Indeed, the authors show that dynamical system strategies based upon nonsmooth gradients have geometric interpretations related to Voronoi diagrams. The authors have modified an already very nicely written original paper to make it even more accessible to a general SIAM audience. We invite readers to enjoy this very important and illuminating view into the world of dynamical systems and optimization. Who knows, one day your car may incorporate technology stemming from this research!

스테레오 기반의 장애물 회피 알고리듬

This paper deals with obstacle avoidance for unmanned vehicle using stereo system. The "DARPA Grand Challenge 2005" shows that the robot can move autonomously under given waypoint. RADAR, IMS (Inertial Measurement System), GPS, camera are used for autonomous navigation. In this paper, we focus on stereo system for autonomous navigation. Our approach is based on Singh et. al. [5]'s approach that is successfully used in an unmanned vehicle and a planetary robot. We propose an improved algorithm for obstacle avoidance by modifying the cost function of Singh et. al. [5]. Proposed algorithm gives more sharp contrast in choosing local path for obstacle avoidance and it is verified in experimental results.

The 2005 DARPA Grand Challenge – The Great Robot Race

The DARPA Grand Challenge was a landmark in the field of robotics: a race by autonomous vehicles through 132 miles of rough, cross-country Nevada terrain that showcased exciting and unprecedented capabilities in robotic perception, navigation, and control. The event took place in October 2005, and drew teams of competitors from academia and industry, and many garage hobbyists. This book presents fifteen technical papers that are written at a level that makes them easily accessible to a broad technical audience, describing the technology behind most of the robotic vehicles that participated in this famous race. The papers describe each team's driverless vehicle, race strategy, and insights. As a whole, they present the state of the art in autonomous vehicle technology, and offer a glimpse of future technology for tomorrows driverless cars. This book will serve as an authoritative, archival source for the DARPA Grand Challenge and a must have for robotics students and researchers, since it describes the state of the art in perception, planning and control.

A perspective on emerging automotive safety applications, derived from lessons learned through participation in the DARPA Grand Challenges

AbstractThis paper reports on various aspects of the Intelligent Vehicle Systems (IVS) team's involvement in the recent 2007 DARPA Urban Challenge, wherein our platform, the autonomous “XAV‐250,'' competed as one of the 11 finalists qualifying for the event. We provide a candid discussion of the hardware and software design process that led to our team's entry, along with lessons learned at this event and derived from participation in the two previous Grand Challenges. In addition, we give an overview of our vision‐, radar‐, and LIDAR‐based perceptual sensing suite, its fusion with a military‐grade inertial navigation package, and the map‐based control and planning architectures used leading up to and during the event. The underlying theme of this article is to elucidate how the development of future automotive safety systems can potentially be accelerated by tackling the technological challenges of autonomous ground vehicle robotics. Of interest, we will discuss how a production manufacturing mindset imposes a unique set of constraints upon approaching the problem and how this worked for and against us, given the very compressed timeline of the contests. © 2008 Wiley Periodicals, Inc.

Robosemantics: How Stanley the Volkswagen Represents the World

One of the most impressive feats in robotics was the 2005 victory by a driverless Volkswagen Touareg in the DARPA Grand Challenge. This paper discusses what can be learned about the nature of representation from the car's successful attempt to navigate the world. We review the hardware and software that it uses to interact with its environment, and describe how these techniques enable it to represent the world. We discuss robosemantics, the meaning of computational structures in robots. We argue that the car constitutes a refutation of semantic arguments against the possibility of strong artificial intelligence.

Managing complexity in an autonomous vehicle

Team Tormenta fielded an autonomous vehicle in the DARPA Grand Challenge 2005. Limiting the complexity of the autonomous system design, construction, and testing was crucial in rapidly designing a cost-effective working system. This approach allowed Team Tormenta to field a competitive vehicle on a low budget competing in the qualification trials.

RoadCompass: following rural roads with vision + ladar using vanishing point tracking

We present a vision- and ladar-based approach to autonomous driving on rural and desert roads that has been tested extensively in a closed-loop system. The vision component uses Gabor wavelet filters for texture analysis to find ruts and tracks from which the road vanishing point can be inferred via Hough-style voting, yielding a direction estimate for steering control. The ladar component projects detected obstacles along the road direction onto the plane of the front of the vehicle and tracks the 1-D obstacle "gap" presumed due to the road to yield a lateral offset estimate. Several image- and state-based tests to detect failure conditions such as off-road poses (i.e., there is no road to follow) and poor lighting due to sun glare or distracting shadows are also explained. The system's efficacy is demonstrated with analysis of diverse logged data including from the 2005 DARPA Grand Challenge, as well as tests with full control of a vehicle over 15 km of difficult roads at up to 37 km/h with no waypoints.

What can be done with an embedded stereo-rig in urban environments?

The development of the Autonomous Guided Vehicles (AGVs) with urban applications are now possible due to the recent solutions (DARPA Grand Challenge) developed to solve the Simultaneous Localization And Mapping (SLAM) problem: perception, path planning and control. For the last decade, the introduction of GPS systems and vision have been allowed the transposition of SLAM methods dedicated to indoor environments to outdoor ones. When the GPS data are unavailable, the current position of the mobile robot can be estimated by the fusion of data from odometer and/or Inertial Navigation System (INS). We detail in this article what can be done with an uncalibrated stereo-rig, when it is embedded in a vehicle which is going through urban roads. The methodology is based on features extracted on planes: we mainly assume the road at the foreground as the plane common to all the urban scenes but other planes like vertical frontages of buildings can be used if the features extracted on the road are not enough relevant. The relative motions of the coplanar features tracked with both cameras allow us to estimate the vehicle ego-motion with a high precision. Futhermore, the features which don’t check the relative motion of the considered plane can be assumed as obstacles.

Software & Systems Engineering Process and Tools for the Development of Autonomous Driving Intelligence

When a large number of people with heterogeneous knowledge and skills run a project together, it is important to use a sensible engineering process. This especially holds for a project building an intelligent autonomously driving car to participate in the 2007 DARPA Urban Challenge. In this article, we present essential elements of a software and systems engineering process for the development of artificial intelligence capable of driving autonomously in complex urban situations. The process includes agile concepts, like test first approach, continuous integration of every software module and a reliable release and configuration management assisted by software tools in integrated development environments. However, the most important ingredients for an efficient and stringent development are the ability to efficiently test the behavior of the developed system in a flexible and modular simulator for urban situations. I. I ntroduction O CUSED research is often centered around interesting challenges and awards. The airplane industry started off with awards for the first flight over the British Channel as well as the Atlantic Ocean. The Human Genome Project, the Robo Cups and the series of DARPA Grand Challenges for autonomous vehicles serve this very same purpose to foster research and development in a particular direction. The 2007 DARPA Urban Challenge 16

Autonomous Driving with Laser Scanners

Team-LUX, a cooperation between Ibeo and Sick, is Ibeo’s first team for the Darpa Grand Challenge race. It is an interesting way of integrating both sensors and software into the driver assistance system for autonomous driving. The answer to the Urban Challenge is the LUX robot, Figure 1.

Driverless cars

The latest DARPA Grand Challenge is all about control in an urban environment, and will be the toughest test yet, explains the author.

Systems for Safety and Autonomous Behavior in Cars: The DARPA Grand Challenge Experience

In this paper, we review technologies for autonomous ground vehicles and their present capabilities in research and in the automotive market. We outline technology requirements for enhanced functions and for infrastructure development. Since the recent Grand Challenge competition is a major force to advance technology in this field, we specifically refer to our experiences in developing a participating vehicle. We present a multisensor platform that has been proven in an off-road environment. It combines different sensing modalities that inherently yield uncertain information. Finite-state machines are formulated to generate rule-based autonomous behavior that enables fully autonomous off-road driving. Overall, the intent of the paper is to evaluate approaches and technologies used in the two Grand Challenges as they contribute to the needs of autonomous cars on the road