OneFast440
Active member
It’s just a suction cup mount that came with my unit.
Sponsored
Sponsored
OpenPilot on the F-150 (BlueCruise alternative)
- Thread starter hitchhiker
- Start date
- Watchers 92
Sam Ellethy
Member
- First Name
- Sam
- Joined
- Sep 13, 2023
- Threads
- 0
- Messages
- 17
- Reaction score
- 12
- Location
- Roswell, Georgia
- Vehicles
- 2023 F-150 Tremor
Thank you, I’ll have to look for mine then ?
Ajzride
Well-known member
Who wants more charts?
Reviving a topic from early in this thread about how ADAS systems handle coming up on stopped traffic. I was doing a lot of testing this morning and ran across one of those scenarios. I was doing 59mph in a 55 zone approaching a car that was completely stopped.
Let's first take a peek at the video:
The truck didn't stop itself, I had to help it. This was utilizing OpenPilot longitudinal controls mixed with Ford Radar. I had previously stated that radar does a poor job of detecting stationary objects, I wondered if the data proved that out. Here's the charts:
Notice the time span on the graph is 3 seconds. If you peek at your phone you miss the whole thing.
The top line is acceleration. At the very beginning it's zero because I'm at a constant speed. Once braking starts you see it go negative (slowing down).
Second line is speed (in meters per second)
Third line is what vision detected. it detects the lead briefly at 107 meters out at about 1.25 seconds into the graph, loses it for a few beats, then picks it back up. It takes 0.04 seconds from the time it detects a lead until the time it starts braking.
The bottom line is what radar picked up... absolutely nothing.
In this case vision picked up a lead at 107m out and started braking for it. OpenPilot braking is limited to regular braking and can't utilize the emergency braking system (safety rules), so it was unable to stop from 59-0 in 107 meters. Putting my foot on the brake gets full braking force and allows me to stop in time.
If this had been an emergency situation where say I blacked out due to low blood sugar, a vision based system would have had a low speed collision resulting in some soreness, but nothing major. The ford radar based system would have hit the car in front of me at 59 mph.
The reason I share this is to make sure everyone has the right expectations for these systems. They are considered Level 2 for a reason. They can't handle all situations and the driver has to be paying attention and ready to intervene, not just for unusual circumstances (like a deer darting across the road) but also for normal circumstances that call for more control than a L2 system can apply.
A vision based system should be able to avoid this completely, but right now we are hardware limited. We don't have enough computing power in a car to solve the model out to 300 or 500 meters. Teslas have the best vision system right now and they have a significant amount of computing power on board, equivalent to a gaming PC. Ford's system and the comma system are closer to cell phone's than gaming computers and they are much more limited on the distance out they can solve the model.
Reviving a topic from early in this thread about how ADAS systems handle coming up on stopped traffic. I was doing a lot of testing this morning and ran across one of those scenarios. I was doing 59mph in a 55 zone approaching a car that was completely stopped.
Let's first take a peek at the video:
The truck didn't stop itself, I had to help it. This was utilizing OpenPilot longitudinal controls mixed with Ford Radar. I had previously stated that radar does a poor job of detecting stationary objects, I wondered if the data proved that out. Here's the charts:
Notice the time span on the graph is 3 seconds. If you peek at your phone you miss the whole thing.
The top line is acceleration. At the very beginning it's zero because I'm at a constant speed. Once braking starts you see it go negative (slowing down).
Second line is speed (in meters per second)
Third line is what vision detected. it detects the lead briefly at 107 meters out at about 1.25 seconds into the graph, loses it for a few beats, then picks it back up. It takes 0.04 seconds from the time it detects a lead until the time it starts braking.
The bottom line is what radar picked up... absolutely nothing.
In this case vision picked up a lead at 107m out and started braking for it. OpenPilot braking is limited to regular braking and can't utilize the emergency braking system (safety rules), so it was unable to stop from 59-0 in 107 meters. Putting my foot on the brake gets full braking force and allows me to stop in time.
If this had been an emergency situation where say I blacked out due to low blood sugar, a vision based system would have had a low speed collision resulting in some soreness, but nothing major. The ford radar based system would have hit the car in front of me at 59 mph.
The reason I share this is to make sure everyone has the right expectations for these systems. They are considered Level 2 for a reason. They can't handle all situations and the driver has to be paying attention and ready to intervene, not just for unusual circumstances (like a deer darting across the road) but also for normal circumstances that call for more control than a L2 system can apply.
A vision based system should be able to avoid this completely, but right now we are hardware limited. We don't have enough computing power in a car to solve the model out to 300 or 500 meters. Teslas have the best vision system right now and they have a significant amount of computing power on board, equivalent to a gaming PC. Ford's system and the comma system are closer to cell phone's than gaming computers and they are much more limited on the distance out they can solve the model.
OneFast440
Active member
Ajzride - Would you be able to send me a message on how to locate the data you are providing? I would love to do the same based on specific scenarios that I run into.
The information you provided goes to show the importance of having the visual ability that Comma provides. Now to get it paired and working together with the radar!
To continue with the importance of paying attention, I had a braking situation yesterday. See the video. While I was prepared to brake in this situation, I actually did not need to intervene, but I was prepared if necessary. The video was only using the Ford stock ACC, not the SunnyPilot Beta Longitudinal controls (which probably would have done better in this situation).
The information you provided goes to show the importance of having the visual ability that Comma provides. Now to get it paired and working together with the radar!
To continue with the importance of paying attention, I had a braking situation yesterday. See the video. While I was prepared to brake in this situation, I actually did not need to intervene, but I was prepared if necessary. The video was only using the Ford stock ACC, not the SunnyPilot Beta Longitudinal controls (which probably would have done better in this situation).
Ajzride
Well-known member
I'll send you a message, no need to muddy this thread with that level of technical stuff.
The current release of SunnyPilot is using radar reinforced vision. The reason we have recommended everyone stay on Ford ACC is not the radar/vision, it is the gas/brake control.
Sponsored
OneFast440
Active member
Thanks for clearing that up.
labjr1
Well-known member
This is great information and an awesome explanation. Intel does make some gaming NUCs that might have the processing power needed to expand the visual range computations, but I suspect the variables get exponentially more complex as distance increases.Who wants more charts?
Reviving a topic from early in this thread about how ADAS systems handle coming up on stopped traffic. I was doing a lot of testing this morning and ran across one of those scenarios. I was doing 59mph in a 55 zone approaching a car that was completely stopped.
Let's first take a peek at the video:
The truck didn't stop itself, I had to help it. This was utilizing OpenPilot longitudinal controls mixed with Ford Radar. I had previously stated that radar does a poor job of detecting stationary objects, I wondered if the data proved that out. Here's the charts:
Notice the time span on the graph is 3 seconds. If you peek at your phone you miss the whole thing.
The top line is acceleration. At the very beginning it's zero because I'm at a constant speed. Once braking starts you see it go negative (slowing down).
Second line is speed (in meters per second)
Third line is what vision detected. it detects the lead briefly at 107 meters out at about 1.25 seconds into the graph, loses it for a few beats, then picks it back up. It takes 0.04 seconds from the time it detects a lead until the time it starts braking.
The bottom line is what radar picked up... absolutely nothing.
In this case vision picked up a lead at 107m out and started braking for it. OpenPilot braking is limited to regular braking and can't utilize the emergency braking system (safety rules), so it was unable to stop from 59-0 in 107 meters. Putting my foot on the brake gets full braking force and allows me to stop in time.
If this had been an emergency situation where say I blacked out due to low blood sugar, a vision based system would have had a low speed collision resulting in some soreness, but nothing major. The ford radar based system would have hit the car in front of me at 59 mph.
The reason I share this is to make sure everyone has the right expectations for these systems. They are considered Level 2 for a reason. They can't handle all situations and the driver has to be paying attention and ready to intervene, not just for unusual circumstances (like a deer darting across the road) but also for normal circumstances that call for more control than a L2 system can apply.
A vision based system should be able to avoid this completely, but right now we are hardware limited. We don't have enough computing power in a car to solve the model out to 300 or 500 meters. Teslas have the best vision system right now and they have a significant amount of computing power on board, equivalent to a gaming PC. Ford's system and the comma system are closer to cell phone's than gaming computers and they are much more limited on the distance out they can solve the model.
Spiffy
Well-known member
This starts to clear up some of my interpretations of each system. I had had situations where someone was running a red light and my truck slammed on the brakes saving from a collision.
Coming off a ramp at 70 with ACC on will not stop in time. So I'm confused at just what ford's collision avoidance does.
Ajzride... a other excellent post !!!!
Coming off a ramp at 70 with ACC on will not stop in time. So I'm confused at just what ford's collision avoidance does.
Ajzride... a other excellent post !!!!
Ajzride
Well-known member
Since Ford doesn't give us the logic, we have to infer it from captured data logs. To really understand it will require triggering the evasive steering and emergency braking multiple times. Since you are trained vehicle tester, I volunteer you to go collect that data.
Spiffy
Well-known member
Trained vehicle tester? Lol.
My testing doesn't involve trees, dear, or stupid drivers. At 10 thousand feet not much to hit. I just need to avoid hitting the dirt. ???
My testing doesn't involve trees, dear, or stupid drivers. At 10 thousand feet not much to hit. I just need to avoid hitting the dirt. ???
Sponsored
Fourthewin
Well-known member
- First Name
- Andrew
- Joined
- Sep 2, 2023
- Threads
- 12
- Messages
- 101
- Reaction score
- 73
- Location
- Kansas City
- Vehicles
- 2021 F-150 Platinum
- Occupation
- Real Estate
Updating to this now.
RobertKingRanch2023
Member
- First Name
- Robert
- Joined
- Dec 24, 2023
- Threads
- 0
- Messages
- 5
- Reaction score
- 4
- Location
- Savannah, Ga
- Vehicles
- 2023 Ford F 150 King Ranch
- Occupation
- Retired Army
It took a while, but with the help of [B]Fourthewin[/B] I got Open Pilot working on my truck (2023 F-150 King Ranch Powerboost). I was at it for two weeks, and was about give up and list it on eBay, when I decided to send a DM to [B]Fourthewin[/B] he replied back and had me fixed up and running in less than an hour. For anyone on the fence, this is the real deal. I have a friend with a 2023 Platinum with B.C. he can only use it on highways, mine was driving the road in my subdivision, Route 17 and I 16 all with no issues, AND no $700.00 subscription Thank you to the trail blazers. You guys really are doing an outstanding job.
Funny thing is I only found this thread because I Googled if I could use an OBD2 port splitter with a dash cam. Well dash cam is in the box next to my desk, I am going to send it to a friend in Tennessee.
The only thing I really don't like about it is that it is attached to the windshield.
I wonder if I can glue it to the mirror . . . . just kidding.
Again thank all you guys who are working so hard at this. A special thank you to [B]Fourthewin[/B] for not only helping set it up last night, and offering some pointers today.
Funny thing is I only found this thread because I Googled if I could use an OBD2 port splitter with a dash cam. Well dash cam is in the box next to my desk, I am going to send it to a friend in Tennessee.
The only thing I really don't like about it is that it is attached to the windshield.
I wonder if I can glue it to the mirror . . . . just kidding.
Again thank all you guys who are working so hard at this. A special thank you to [B]Fourthewin[/B] for not only helping set it up last night, and offering some pointers today.
Fourthewin
Well-known member
- First Name
- Andrew
- Joined
- Sep 2, 2023
- Threads
- 12
- Messages
- 101
- Reaction score
- 73
- Location
- Kansas City
- Vehicles
- 2021 F-150 Platinum
- Occupation
- Real Estate
@RobertKingRanch2023 quite happy to help! The more people using the device the more likely it gets pushed to fully compatible though the 2024 encryption on the new models might interrupt this wish.
It’s great technology and I enjoy it everyday. Quite a few people helped me early including @rwxr_xr_x and @Ajzride, they were the real trailblazers.
It’s great technology and I enjoy it everyday. Quite a few people helped me early including @rwxr_xr_x and @Ajzride, they were the real trailblazers.
Ajzride
Well-known member
It's like The Force guided you to us. It knew what you really wanted and provided.
I do have a prototype of a review mirror mount, but I haven't had a chance to test it to see how much vibration it picks up, it might be unusable.
Please don't fight with something for weeks again. Just drop a post in here and someone will help you out. We even have a how-to doc that should make things much easier (I believe @Fourthewin is adding some clarifications based on what he saw helping you) we could have sent you. Several of us also provide FaceTime/zoom help when needed.
RobertKingRanch2023
Member
- First Name
- Robert
- Joined
- Dec 24, 2023
- Threads
- 0
- Messages
- 5
- Reaction score
- 4
- Location
- Savannah, Ga
- Vehicles
- 2023 Ford F 150 King Ranch
- Occupation
- Retired Army
Call it foolish pride I guess. I have read almost every line in all 28 pages, copy and pasted every URL I could find to a word document, then printed it out and off to the truck I would go. I didn't have the 8° mount yet so I would put it on the dash type in the lines, and get a thousand dollar dash cam. So yeah I should have reach out sooner. Just a another reason for the guys and girls out there who are thinking about it, worried how it works, to just jump in these guys will help. This is not one of those testosterone driven threads, where every answer to every question is a cut down or insult. These guys help. Let me know I would be interested in hearing how that mirror mount workes out. I'd be glad to drive around with it and give you feed back.
Sponsored
