Testing Automatic Emergency Braking on The Mustang Mach E

[superdave80]

I think this video shows perfectly how autonomous is just not a 100% solution same as a human and a combination of both is probably the best

Any system that has the vehicle steer/navigate the car will never be able to be merged with an active/attentive driver because a human will (over time) pay less and less attention as they get use to the vehicle driving itself.

Sponsored

 

[i8iridium]

Agreed.


That's the main issue with the tests. The AEB needs to have something to differentiate from the road and objects while not having false positives. So it's going to err on the side of hitting a pedestrian vs slamming on the brakes for a manhole cover or shadows. It might work better (probably not great) if you had someone pull something across the road with a rope while you drove at it.

There's only a single camera up front, so it has a lot of work to do to pick out a person at speed. From what I understand, the radar isn't even being used unless you're at higher speed. I've had that work reliably when someone's turned off in front of me, and I didn't hit the brakes cause I knew I would clear them. Car still applied the brakes anyway and I would just have to hit the accelerator.

I've had the reverse braking work 100% with a regular kitchen garbage can behind the car. ? Again, that's only using a single camera.

Here's some reading material to supplement/correct some of my post and show the flaws in the system and the testing in the video.

Pre-Collision Assist System Operation


The pre-collision assist system is safety feature on vehicles equipped with ACC . It uses both the CCM (radar) and the IPMA . The system is active whether the ACC system is on or off. If the system detects a vehicle, pedestrian or other object in the vehicle path of travel, the system provides three levels of functionality:

• Visual and audible alert
• Brake support
• Active braking

Vehicles equipped with ACC use object detection information from the radar sensor integrated in the CCM and the forward-looking camera in the IPMA mounted on the windshield below the rear view mirror. The CCM and the IPMA scan a designated area in front of the vehicle. Network messages sent between the CCM and the IPMA on dedicated CAN circuits, which determine whether an object, vehicle or pedestrian is in the path of travel, the approximate distance to the object and how fast the vehicle is approaching it.

When the vehicle approaches the object, the IPMA sends a message through the GWM to the IPC module to turn on the red pre-collision warning indicator in the message center and to sound an audible alert. The IPMA also sends a message to the ABS module to pre-charge the brakes in order to prepare them for rapid braking. If the system determines that a collision is imminent, an active braking message is sent from the IPMA to the ABS module. The system reduces the gap between the brake pads and discs, applying the brakes to slow vehicle speed without driver intervention.


The pre-collision alert system uses image recognition software that differentiates shapes, which allows the system to determine if the approaching object is a vehicle or a pedestrian. If the camera does not recognize the shape as a vehicle or a pedestrian, the system does not provide full function.


The system may not work properly at night, in direct or low sunlight, when camera vision is reduced due to weather conditions or a blocked CCM radar sensor. If the IPMA camera module is obstructed, the pre-collision system does not respond properly to pedestrians or stationary vehicles and reduces the ability to recognize moving vehicles. Unconventional vehicle types, pedestrians in groups or with complex backgrounds or partly obscured pedestrians may not be detected by the system.


The pre-collision assist system is active at speeds above 5 km/h (3 mph) and pedestrian detection is active at speeds up to 80 km/h (50 mph).


The pre-collision assist system has sensitivity detection that can be changed through the steering wheel switch. The alert sensitivity is adjusted and active braking can be turned off in the IPC . The active braking function reverts back to ON at the next ignition cycle.


When a system fault is detected with the pre-collision warning system, the message PRE-COLLISION ASSIST NOT AVAILABLE SENSOR BLOCKED or PRE-COLLISION ASSIST NOT AVAILABLE is displayed in the IPC message center.

 

[Zigma51]

Agreed.


That's the main issue with the tests. The AEB needs to have something to differentiate from the road and objects while not having false positives. So it's going to err on the side of hitting a pedestrian vs slamming on the brakes for a manhole cover or shadows. It might work better (probably not great) if you had someone pull something across the road with a rope while you drove at it.

There's only a single camera up front, so it has a lot of work to do to pick out a person at speed. From what I understand, the radar isn't even being used unless you're at higher speed. I've had that work reliably when someone's turned off in front of me, and I didn't hit the brakes cause I knew I would clear them. Car still applied the brakes anyway and I would just have to hit the accelerator.

I've had the reverse braking work 100% with a regular kitchen garbage can behind the car. ? Again, that's only using a single camera.

Definitely good information for part 2

 

[Blue highway]

In this video I use cute punching bags to test the automatic emergency braking on the Mustang Mach E.


Don't forget to follow me on Instagram: https://www.instagram.com/mach_e_connection/

If you have any questions, comments, or challenges, please post below.

Inflatable animals have Zero radar cross sectional area, and they are not moving The ultrasonic sensor can detect them at low speeds but at speeds that depend on Radar they are problematic because they are invisible to the radar.

If you did this with a metal trash can, you might get a different result but I doubt it unless the trashcans were moving. For the collision detection system to work, the objects have to be visible to the camera and or radar, and for most systems, they have to be moving. Otherwise they look just like a speed sign or a metal over pass (lots of radar reflection, but no motion) which are ignored because there is no relative motion (different than the speed of the car). If they weren't ignored the car would slam on the brakes for every low overpass or speed sign on a curve.

 

[Jimbo]

I challenge you to post content here that doesn't include revenue-generating videos.

Kinda rich coming from the guy who spams Amazon affiliate links in every one of his posts

 

[Mach-Lee]

Uh, sorry but your test is flawed. The cameras in the vehicle are trained to recognize humans and cars, not blow up toys. The Mach-E AEB has been tested by several agencies:





The night time tests are no joke. The dummies they use are made with special radar-reflecting foam and use movable limbs. The camera looks for the shape of the human and whether it is moving into the path of the vehicle. It will happily crash into a blowup toy or a trash can because it is not a human. The fact remains the Mach-E has some of the best advanced driver assistance and collision avoidance technology on the market.

Try your reverse test with 1PD off and just coasting (feet not touching any pedals). You might have to simulate movement by dragging or rolling an object into the path of the vehicle.

 

[Zigma51]

Uh, sorry but your test is flawed. The cameras in the vehicle are trained to recognize humans and cars, not blow up toys. The Mach-E AEB has been tested by several agencies:





The night time tests are no joke. The dummies they use are made with special radar-reflecting foam and use movable limbs. The camera looks for the shape of the human and whether it is moving into the path of the vehicle. It will happily crash into a blowup toy or a trash can because it is not a human. The fact remains the Mach-E has some of the best advanced driver assistance and collision avoidance technology on the market.

Try your reverse test with 1PD off and just coasting (feet not touching any pedals). You might have to simulate movement by dragging or rolling an object into the path of the vehicle.

Good information for part 2, thank you.

 

[21st Century Pony]

I appreciate the test video - thank you for posting it.

While watching it, I also was thinking about the density and the reflectivity of the test dummy test dolphin.

How about this: a normal-size plastic garbage can, on a rope, dragged by another car down the cul-de-sac at a slow speed (slower than the Mach E approach speed). Will the Mach E break automatically?

If I wasn't all the way across the country, I might have volunteered here to be the dummy-pulling car. Just for benefit of science, ya know?

 

[Zigma51]

Not sure I want to run into a plastic garbage can, that can do some damge I feel. I have some ideas for part 2 that might be better.

 

[Mach-Lee]

How about this: a normal-size plastic garbage can, on a rope, dragged by another car down the cul-de-sac at a slow speed (slower than the Mach E approach speed). Will the Mach E break automatically?

Nope, because it doesn't look like a person walking.

 

[21st Century Pony]

Nope, because it doesn't look like a person walking.

It might look like a person staggering

 

[Zigma51]

Nope, because it doesn't look like a person walking.

On the backup it did see the dolphin cause it was beeping. On the slow drive tests Tesla also has issues here
But I have some ideas to safely try this again and try some things brought up in this forum. Don't plan to do part 2 until after blue cruise 1.2 update

 

[Mach-Lee]

On the backup it did see the dolphin cause it was beeping. On the slow drive tests Tesla also has issues here
But I have some ideas to safely try this again and try some things brought up in this forum. Don't plan to do part 2 until after blue cruise 1.2 update

FYI that Dawn project video has been debunked as fake. They deliberately did not engage FSD properly/fully.

https://electrek.co/2022/08/10/tesl...mpaign-releases-test-fails-fsd-never-engaged/

 

[Zigma51]

FYI that Dawn project video has been debunked as fake. They deliberately did not engage FSD properly/fully.

https://electrek.co/2022/08/10/tesl...mpaign-releases-test-fails-fsd-never-engaged/

Oh did not know that. That relly shitty

 

[axelheimer24]

Every time you kill the blow up animal I laughed, it doesn't prove a thing but it really was entertaining. I think you need much bigger objects for it to be a real test. Perhaps One of those big blow up dancing men.

Let the government agencies do the “realistic” tests. Please please please post a follow up video of your Mach e driving over a wacky wavy inflatable tube man at various speeds! I’m personally hoping there is an Easter egg in the software that makes sound effects upon impact.

Sponsored