Overtake Scenario

Even if this Scenario is quite simple, in our opinion is probably the most important Scenario, as it is the one we used to test the Autonomous Overtaking Path Planning as well as the Decision Making State Machine: in fact, this was the first Scenario which we used for the autonomous Overtaking and all the following ones were built onto this to check that further complicating the Scenario would not damage the good working of our System.

The first thing which we want to comment is the trend of the State during the

Figure 3.43: State during the Overtake Scenario

simulation: differently from what we saw in previous Scenarios - where the State was constantly in 1 (STAY) or eventually 2 (WAIT) - here we see that the State jumps into 3 (OVERTAKE), as we would expect and after some seconds moves to 6 (GO BACK) once the leading vehicle is overtaken. Eventually, the State settles into 1 (STAY) after the overtake is completed and the EgoVehicle is back in the OwnLane.

We can also see from Figure 3.43 that the State jumps to 3 from basically the first second of the simulation: this is because the leading vehicle is immediately in range of our sensors and this is confirmed by the trend of the Leading Flag in Figure 3.44 which is raised immediately.

Since the Scenario was supposed to be a simple scenario just to test the Autonomous Overtaking Path Planning and the Decision Making, we decided not to put any oncoming vehicle, so that we were not bothered by the OvtCounter (Section 2.7.1) and by sudden Abort maneuvers or Emergency Overtakes: this can be confirmed

Figure 3.44: Leading Flag during the Overtake Scenario

by looking at Figure 3.45, which depicts an Oncoming Flag constantly equal to 0.

Another important thing we could comment are the signals of the Radar, meaning

Figure 3.45: Oncoming Flag during the Overtake Scenario

the relative distance ∆X and ∆V relative speed, which are represented in Figure 3.46: we can see that the relative distance is steadily dropping while the relative speed is not changing a lot: this is not due to the leading vehicle inreasing its speed,

but simply to the fact that, since we are in State 3 (OVERTAKE) and not in State 4 (EMERGENCY OVT), the EgoVehicle is not forced to accelerate as much as possible, but tries to keep the set speed Vdes, as we can also see from Figure 3.48

Figure 3.46: Radar signals during the Overtake Scenario

Figure 3.47: Time gap during the Overtake Scenario

where the initial drop is as always due to the starting acc command of 0: because

Figure 3.48: Speed of the EgoVehicle during the Overtake Scenario

of this, the ∆V is dropping through all of the time represented in Figure 3.46: this is because the Radar is tracking a leading vehicle only for the first 4-5 seconds, which is the portion of time in which the EgoVehicle speed drops; this limited time of the leading vehicle tracking explains why the Leading Flag duration is much shorter than the State 3 OVERTAKING, as we can see by comparing with the e_h which is an indication of our yaw angle ψ: the peak up to 7° of eh is due to the car heavily steering towards the left, therefore leading the camera to see the Lane Boundaries going towards the right. The most important consequence of the vehicle steering, however is that the leading vehicle exits from the Field of View of the Central Radar and, therefore, the Leading Flag falls.

The heading error peak is, however, something which we cannot avoid, since the vehicle is not equipped with four wheel steering (which would allow to change lane without yawing) and we want to indeed change the lane to perform the Overtake.

The good part is that the eh rapidly falls to 0, meaning that the EgoVehicle quickly gets stabilized in the Oncoming Lane. Around the 10 s mark we see a negative spike in the eh: this means that the Overtaking has been completed and we are now steering right to go back in the OwnLane. This is confirmed by the fact that at around the same 10 s mark, the State passes from 3 to 6 (GO BACK).

The Cross-Track Error displays the same trend as the Heading Error, even though by looking at the graphs of Figures 3.49 and 3.50 we could be misled into thinking the opposite. However, the sharp fall from 1.5 m to -1.3 is due to the fact that the EgoVehicle is now in the Oncoming Lane, even though just on the right boundary of said Lane; because of this, we move from being on the very left boundary of the

Figure 3.49: Heading Error during the Overtake Scenario

Figure 3.50: Cross-Track Error during the Overtake Scenario

Own Lane (causing ect = +1.5m) to being on theb right boundary of the Oncoming Lane (which should be causing ect= −1.5m, but is causing a local minimum around -1.3, due to the fact that the car is not parallel to the boundary lanes, leading to

some error on the ect perception.

As a confirmation of this explanation, we see that the error on the Go Back

movement (around the 13 s mark), which is reversed, so first a negative error and then a positive one, is much closer to 1.5 m in both the peaks, since the Heading Angle is less than half in the second Lane Change. To conclude this Section, we

Figure 3.51: Steering Command during the Overtake Scenario

are going to present the trajectory followed by the EgoVehicle during this Scenario run: the Vehicle performs a sharp turn to the left in just 100 m, then stays in the Oncoming Lane and begins the Go Back maneuver just after 200 m in total; from Figure 3.53 we can see that the EgoVehicle is back in the Own Lane after just 260 m, traveling a total of just 170 m in the Oncoming Lane. Even though this is of course caused by the large difference in speed between us and the leading vehicle, but we can state that the first Autonomous Overtaking went well and so we moved to more complicated Scenarios.

Figure 3.52: Trajectory followed during the Overtake Scenario

Figure 3.53: Zoom of the trajectory followed during the Overtake Scenario