<p>This thesis project was carried out at Volvo Car Corporation. It is based on an EU project called Locust in which a bio-inspired visual sensor system (the Locust sensor system) for automotive collision avoidance was developed. The Locust sensor system is designed to emulate the collision avoidance functionality of the Locust grasshopper, which is well-known for its extraordinary vision based collision avoidance ability, in particular with regard to its fast reaction times to perceived threats. Volvo Car Corporation is interested in the possibility of using the bio-inspired technology developed in the Locust project to improve its already existing collision avoidance systems. Pedestrian collision avoidance is of high interest, for which the properties of the Locust grasshopper are desirable.</p><p>The purpose of this thesis project is to develop two demonstrator vehicles to test the performance of the Locust sensor system, carry out the testing, and evaluate its usability for Volvo Car Corporation. The first vehicle is a scale 1:5 model car that was originally developed in a thesis project at KTH, and the second a full scale Volvo XC90.</p><p>It was found in the testing that the Locust sensor system is promising for pedestrian collision avoidance applications. The results for detecting other vehicles were also acceptable, but Volvo Car Corporation already has other collision avoidance systems with better performance in this regard. In general the test results were very good for speeds up to about 40 km/h. This indicates that the Locust sensor system would be most usable in a city driving environment, parking lot situations, and for driving in residential areas.</p>

<p>Intervjuer med kunniga på AIS-området har utförts parallellt med en enkät riktad till aktiva befäl ute till sjöss. Detta i syfte att ta reda på dels hur detta hjälpmedel bör användas vad avser antikollision, dels hur det faktiskt används i praktiken.</p><p>AIS, Automatic Identification System, är ett informationsdelningssystem som hjälper fartyg att navigera säkert genom att sända ut egen navigationsdata och samtidigt ta emot motsvarande information från andra fartyg. Systemet nyttjas även som informationssystem för landbaserade stationer såsom VTS-centraler och kustbevakning.</p><p>Undersökningen har engagerat och visat att expertisen anser att AIS, i många avseenden, är ett bra och användbart hjälpmedel som <em>bör</em> komplettera övrig utrustning vid bedömning om kollisionsrisk. Vidare visar enkäten att ARPA-systemet, trots vissa inneboende brister, tillmäts större förtroende än AIS och AIS-vektorer används därför sparsamt. Enkäten visar även att en stor del av de aktiva inte är främmande för att göra avsteg från sjövägsreglerna emedan expertisen i huvudsak avråder från sådant.</p> / <p>Interviews with experts, in the field of AIS, have been conducted alongside a quantitative survey aimed at certified seagoing officers. The purpose of the survey was to investigate how this navigational aid preferably should be used in collision avoidance, in contrast to how it is used in reality.</p><p>AIS, Automatic Identification System, is an information sharing system to assist in safe navigation by transmitting the own vessels navigational data while, at the same time, receiving corresponding information from other ships. The system is also utilized by land based stations such as VTS-centers and coast guards.</p><p>The investigation shows that experts conclude AIS to be, in many aspects, a good and useful aid which <em>should</em> be used in collaboration with other equipment when assessing risk of collision. Furthermore, the survey show that the ARPA-system, in spite of some internal shortcomings, is awarded more trust than AIS and AIS vectors are therefore rarely used. The survey also shows that a large number of officers are open to deviating from the COLREGs while the experts, in general, advise against it.</p>

This paper explores the motion planning problem for multiple moving objects. The approach taken consists of assigning priorities to the objects, then planning motions one object at a time. For each moving object, the planner constructs a configuration space-time that represents the time-varying constraints imposed on the moving object by the other moving and stationary objects. The planner represents this space-time approximately, using two-dimensional slices. The space-time is then searched for a collision-free path. The paper demonstrates this approach in two domains. One domain consists of translating planar objects; the other domain consists of two-link planar articulated arms.

This thesis presents a 3D reconstruction approach to the detection of static obstacles from a single rear view parking camera. Corner features are tracked to estimate the vehicle’s motion and to perform multiview triangulation in order to reconstruct the scene. We model the camera motion as planar motion and use the knowledge of the camera pose to efficiently solve motion parameters. Based on the observed motion, we selected snapshots from which the scene is reconstructed. These snapshots guarantee a sufficient baseline between the images and result in more robust scene modeling. Multiview triangulation of a feature is performed only if the feature obeys the epipolar constraint. Triangulated features are semantically labelled according to their 3D location. Obstacle features are spatially clustered to reduce false detections. Finally, the distance to the nearest obstacle cluster is reported to the driver.

This thesis presents a 3D reconstruction approach to the detection of static obstacles from a single rear view parking camera. Corner features are tracked to estimate the vehicle’s motion and to perform multiview triangulation in order to reconstruct the scene. We model the camera motion as planar motion and use the knowledge of the camera pose to efficiently solve motion parameters. Based on the observed motion, we selected snapshots from which the scene is reconstructed. These snapshots guarantee a sufficient baseline between the images and result in more robust scene modeling. Multiview triangulation of a feature is performed only if the feature obeys the epipolar constraint. Triangulated features are semantically labelled according to their 3D location. Obstacle features are spatially clustered to reduce false detections. Finally, the distance to the nearest obstacle cluster is reported to the driver.

In the last decade, 802.11 wireless devices data-rates have increased by three orders of magnitude, while communications experiencing low throughput are still largely present. Such throughput loss is a fundamental problem of wireless networking that is difficult to diagnose and amend. My research addresses two key causes of throughput loss: MAC layer protocol overhead and destructive link interference. First, I design WiFi-Nano reducing the channel access overhead by an order of magnitude leveraging an innovative speculative technique to transmit preambles. This new concept is based on simultaneous preamble transmission and detection via a self-interference cancellation design, and paves the way to the realization of the collision detection paradigm in wireless networks. Next, I propose 802.11ec (Encoded Control), the first 802.11-based protocol that eliminates the overhead of control packets. Instead, 802.11ec coordinates node transmissions via a set of predefined pseudo-noise codewords, resulting in the dramatic increase of throughput and communication robustness. Finally, I design MIDAS, a model-driven network management tool that alleviates low throughput wireless links identifying key corrective actions. MIDAS' key contribution is to reveal the fundamental role of node transmission coordination in characterizing destructive interference. I implement WiFi-Nano, 802.11ec, and MIDAS using a combination of WARP FPGA-based radio boards, custom emulation platforms, and network simulators. The results obtained show that WiFi-Nano increases the network throughput by up to 100%, 802.11ec improves network access fairness by up to 90%, and MIDAS identifies corrective actions with a prediction error as low as 20%.

This thesis investigates the crucial problem of collision avoidance for autonomous vehicles.  An anti-collision system for an unmanned aerial vehicle (UAV) is studied in particular. The purpose of this system is to make sure that the own vehicle avoids collision with other aircraft in mid-air. The sensor used to track any possible threat is for a UAV limited basically to a digital video camera. This sensor can only measure the direction to an intruding vehicle, not the range, and is therefore denoted an angle-only sensor. To estimate the position and velocity of the intruder a tracking system, based on an extended Kalman filter, is used. State estimates supplied by this system are very uncertain due to the difficulties of angle-only tracking. Probabilistic methods are therefore required for risk calculation. The risk assessment module is one of the essential parts of the collision avoidance system and has the purpose of continuously evaluating the risk for collision. To do this in a probabilistic way, it is necessary to assume a probability distribution for the tracking system output. A common approach is to assume normality, more out of habit than on actual grounds. This thesis investigates the normality assumption, and it is found that the tracking output rapidly converge towards a good normal distribution approximation. The thesis furthermore investigates the actual risk assessment module to find out how the collision risk should be determined. The traditional way to do this is to focus on a critical time point (time of closest point of approach, time of maximum collision risk etc.). A recently proposed alternative is to evaluate the risk over a horizon of time. The difference between these two concepts is evaluated. An approximate computational method for integrated risk, suitable for real-time implementations, is also validated. It is shown that the risk seen over a horizon of time is much more robust to estimation accuracy than the risk from a critical time point. The integrated risk also gives a more intuitively correct result, which makes it possible to implement the risk assessment module with a direct connection to specified aviation safety rules.

A CMbB system is a system that with the help of sensors in the front of a car detects when a collision in unavoidable. When a situation like that is detected, the brakes are activated. The decision of whether to activate the brakes or not is taken by a piece of software called a decision maker. This software continuously checks for routes that would avoid an object in front of the car and as long as a path is found nothing is done. Volvo has been investigating several different CMbB-systems, and the research done by Volvo has previously focused on decision makers that only consider one object in front of the car. By instead taking all present objects in consideration, it should be possible to detect an imminent collision earlier. Volvo has developed some prototypes but needed help evaluating their performance. As part of this thesis a testing method was developed. The idea was to test as many cases as possible but as the objects’ possible states increase, the number of test cases quickly becomes huge. Different ways of removing irrelevant test cases were developed and when these ideas were realized in a test bench, it showed that about 98 % of the test cases could be removed. The test results showed that there is clearly an advantage to consider many objects if the cost of increased complexity in the decision maker is not too big. However, the risk of false alarms is high with the current decision makers and several possible improvements have therefore been suggested.

This thesis project was carried out at Volvo Car Corporation. It is based on an EU project called Locust in which a bio-inspired visual sensor system (the Locust sensor system) for automotive collision avoidance was developed. The Locust sensor system is designed to emulate the collision avoidance functionality of the Locust grasshopper, which is well-known for its extraordinary vision based collision avoidance ability, in particular with regard to its fast reaction times to perceived threats. Volvo Car Corporation is interested in the possibility of using the bio-inspired technology developed in the Locust project to improve its already existing collision avoidance systems. Pedestrian collision avoidance is of high interest, for which the properties of the Locust grasshopper are desirable. The purpose of this thesis project is to develop two demonstrator vehicles to test the performance of the Locust sensor system, carry out the testing, and evaluate its usability for Volvo Car Corporation. The first vehicle is a scale 1:5 model car that was originally developed in a thesis project at KTH, and the second a full scale Volvo XC90. It was found in the testing that the Locust sensor system is promising for pedestrian collision avoidance applications. The results for detecting other vehicles were also acceptable, but Volvo Car Corporation already has other collision avoidance systems with better performance in this regard. In general the test results were very good for speeds up to about 40 km/h. This indicates that the Locust sensor system would be most usable in a city driving environment, parking lot situations, and for driving in residential areas.

Intelligent systems with even the bare minimum of sophistication require extensive computational power and complex processing units. At the same time, small insects like flies are adept at visual navigation, target pursuit, motionless hovering flight, and obstacle avoidance. Thus, biology provides engineers with an unconventional approach to solve complicated engineering design problems. Computational models of the neuronal architecture of the insect brain can provide algorithms for the development of software and hardware to accomplish sophisticated visual navigation tasks. In this research, we investigate biologically-inspired collision avoidance models primarily based on visual motion. We first present a comparative analysis of two leading collision avoidance models hypothesized in the insect brain. The models are simulated and mathematically analyzed for collision and non-collision scenarios. Based on this analysis it is proposed that along with the motion information, an estimate of distance from the obstacle is also required to reliably avoid collisions. We present models with tracking capability as solutions to this problem and show that tracking indirectly computes a measure of distance. We present a camera-based implementation of the collision avoidance models with tracking. The camera-based system was tested for collision and non-collision scenarios to verify our simulation claims that tracking improves collision avoidance. Next, we present a direct approach to estimate the distance from an obstacle by utilizing non-directional speed. We describe two simplified non-directional speed estimation models: the non-directional multiplication (ND-M) sensor, and the non-directional summation (ND-S) sensor. We also analyze the mathematical basis of their speed sensitivity. An analog VLSI chip was designed and fabricated to implement these models in silicon. The chip was fabricated in a 0.18 um process and its characterization results are reported here. As future work, the tracking algorithm and the collision avoidance models may be implemented as a sensor chip and used for autonomous navigation by intelligent systems.