Fig 8 - uploaded by Gabriel Zachmann
Content may be subject to copyright.
Sensor data displayed in the remote control center. LiDAR scans are rendered as point clouds, cameras are mapped to planes.
Context in source publication
Context 1
... remote driving functionality is developed as a backup solution for unexpected scenarios and for monitoring. For this, sensor and preprocessed data are transferred to a control center to be visualized on a standard display or in virtual reality, as shown in Fig. 8. With this an operator will be able to interact with the vehicle directly (in the form of control signals) or indirectly (in the form of waypoints for the car to follow by ...
Similar publications
The underlying mechanisms and physics of catalytic Janus microswimmers is highly complex, requiring details of the associated phoretic fields and the physiochemical properties of catalyst, particle, boundaries, and the fuel used. Therefore, developing a minimal (and more general) model capable of capturing the overall dynamics of these autonomous p...
Multi-camera depth estimation has recently garnered significant attention due to its substantial practical implications in the realm of autonomous driving. In this paper, we delve into the task of self-supervised multi-camera depth estimation and propose an innovative framework, STViT, featuring several noteworthy enhancements: 1) we propose a Spat...
Citations
... In particular, the ⊞-operator is used to generate the sigma points in (37), the measurement function h : S → Z in (38) is defined on manifold spaces, and the iterative algorithm of Table 1 is applied to compute the expected measurement in (39). Furthermore, the ⊟-operator is utilized to determine the difference between the sigma points and the expected measurement in the QR decomposition in (40) and in the Cholesky up-or downdate in (41). The same applies to the calculation of the cross-covariance in (42), where the ⊟-operator is used as well. ...
... The weights of the sample points in (35), (40), (42), and (49) are changed to w i = 1 2n , i = 1, . . . , 2n. ...
... • Since there is no 0-th sample point anymore, the sum in (42) becomes ∑ 2n i=1 , i.e., its lower bound is changed from 0 to 1. • There is no weight anymore that could have a negative value, and all sample points are already considered in the QR decompositions in (35), (40), and (49). Accordingly, the Cholesky up-or downdates in (36) Furthermore, the MEANOFSIGMAPOINTS algorithm in Table 1 needs to be adapted as well to match the changes to the sample points: ...
Estimating the state of a system by fusing sensor data is a major prerequisite in many applications. When the state is time-variant, derivatives of the Kalman filter are a popular choice for solving that task. Two variants are the square-root unscented Kalman filter (SRUKF) and the square-root cubature Kalman filter (SCKF). In contrast to the unscented Kalman filter (UKF) and the cubature Kalman filter (CKF), they do not operate on the covariance matrix but on its square root. In this work, we modify the SRUKF and the SCKF for use on manifolds. This is particularly relevant for many state estimation problems when, for example, an orientation is part of a state or a measurement. In contrast to other approaches, our solution is both generic and mathematically coherent. It has the same theoretical complexity as the UKF and CKF on manifolds, but we show that the practical implementation can be faster. Furthermore, it gains the improved numerical properties of the classical SRUKF and SCKF. We compare the SRUKF and the SCKF on manifolds to the UKF and the CKF on manifolds, using the example of odometry estimation for an autonomous car. It is demonstrated that all algorithms have the same localization performance, but our SRUKF and SCKF have lower computational demands.
... The implemented code related to the RL aspect is based on OpenAI's Spinning Up [35]. The entire trajectory planning and control algorithm is implemented in C++ with the ADTF middleware while the RL algorithm is developed in python using the TensorFlow platform [30]. ...
Deep reinforcement learning (DRL) allows a system to interact with its environment and take actions by training an efficient policy that maximizes self-defined rewards. In autonomous driving, it can be used as a strategy for high-level decision making, whereas low-level algorithms such as the hybrid A* path planning have proven their ability to solve the local trajectory planning problem. In this work, we combine these two methods where the DRL makes high-level decisions such as lane change commands. After obtaining the lane change command, the hybrid A* planner is able to generate a collision-free trajectory to be executed by a model predictive controller (MPC). In addition, the DRL algorithm is able to keep the lane change command consistent within a chosen time-period. Traffic rules are implemented using linear temporal logic (LTL), which is then utilized as a reward function in DRL. Furthermore, we validate the proposed method on a real system to demonstrate its feasibility from simulation to implementation on real hardware.
... based on the research vehicle described in [36]. Thereby, l describes the wheelbase, and the other constants model the physical limitations of the car in normal driving mode. ...
In the realm of autonomous driving, ensuring a secure halt is imperative across diverse scenarios, ranging from routine stops at traffic lights to critical situations involving detected system boundaries of crucial modules. This article presents a novel methodology for swiftly calculating safe stop trajectories. We utilize a clustering method to categorize lane shapes to assign encountered traffic situations at runtime to a set of precomputed resources. Among these resources, there are precalculated halt trajectories along representative lane centers that serve as parametrizations of the optimal control problem. At runtime, the current road settings are identified, and the respective precomputed trajectory is selected and then adjusted to fit the present situation. Here, the perceived lane center is considered a change in the parameters of the optimal control problem. Thus, techniques based on parametric sensitivity analysis can be employed, such as the low-cost feasibility correction. This approach covers a substantial number of lane shapes and exhibits a similar solution quality as a re-optimization to generate a trajectory while demanding only a fraction of the computation time.
... Das OPA 3 L-Projekt an der Universität Bremen setzt ein Hybridfahrzeug zur Erforschung urbaner autonomer Fahranwendungen ein [10]. Dieses Forschungsfahrzeug ist mit einer Vielzahl von Sensoren ausgestattet. ...
... [1], [2]. Tests of MPC implementations in real vehicles have been presented e.g. by [3], [4], [5]. ...
This paper investigates the task of lateral disturbance
compensation based on model predictive control (MPC)
for autonomous vehicles. By considering external disturbances
and parameter perturbations in the model term of the MPC,
the steady-state offset can be compensated. However, in the
presence of more dynamic disturbances, like side wind, the
lateral path tracking performance deteriorates. To overcome
this limitation, a cascaded approach is presented, which is
a combination of an MPC-based and an underlying direct
compensation. The performance of this approach is validated
in simulations as well as in practice with real vehicle tests.
In the joint research project PRORETA 5, building blocks for automated driving in urban areas have been developed, implemented, and tested. The developed blocks involve an object tracking for cars, bicycles, and pedestrians that feeds a multimodal object prediction which is able to predict the traffic participants’ most likely trajectories. Then, an anytime tree-based planning algorithm calculates the vehicle’s desired path. Finally, logic-based safety functions ensure a collision-free trajectory for the ego vehicle. The mentioned building blocks were integrated and tested in a prototype vehicle in urban scenarios. Furthermore, a novel general framework for specifying and testing traffic rule compliance has been developed. In this paper, the automated driving concept of PRORETA 5 is introduced and the developed methods are briefly explained.