# Multiple ArkFlow/Optical Flow

**URL:** <https://discuss.px4.io/t/multiple-arkflow-optical-flow/43083>\
**Category:** PX4 Autopilot\
**Created:** [December 31, 2024, 6:33pm UTC](https://discuss.px4.io/t/multiple-arkflow-optical-flow/43083 "2024-12-31T18:33:17Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![AWilkins\_Ascend](https://discuss.px4.io/user_avatar/discuss.px4.io/awilkins_ascend/32/8388_2.png) [@AWilkins\_Ascend](https://discuss.px4.io/u/AWilkins_Ascend)\
**Post date:** [December 31, 2024, 6:33pm UTC](https://discuss.px4.io/t/multiple-arkflow-optical-flow/43083/1 "2024-12-31T18:33:17Z")

</div>

Was curious if there is any thought/movement toward multi-optical flow for added position hold in environments where downward facing optics might be obscure (say over water or if an optical flow sensor gets blinded with mud).

I imagine that this would take an efk2 change as well as PX4 supporting multiple lidars that wont be used for alt sensing.

Was just curious. Thanks!

 ![image](https://discuss.px4.io/uploads/default/original/3X/c/a/ca9342db2884b2a5f4585a4666a243ba7e9a2e8f.jpeg)

---

<div class="post-metadata">

**Author:** ![dagar](https://discuss.px4.io/user_avatar/discuss.px4.io/dagar/32/22_2.png) [@dagar](https://discuss.px4.io/u/dagar)\
**Post date:** [January 6, 2025, 4:58pm UTC](https://discuss.px4.io/t/multiple-arkflow-optical-flow/43083/2 "2025-01-06T16:58:05Z")

</div>

I’ve done this with multiple instances of ekf2 (each using a different flow sensor) and it might help with redundancy edge cases, but otherwise doesn’t provide much value.

What I’d actually like to try is adding it properly as a new dedicated aid source in the EKF backend (ie optical flow 1, optical flow 2). That way you have additional corrections from each and the ability to catch failures independently.

---

<div class="post-metadata">

**Author:** ![ronron](https://discuss.px4.io/user_avatar/discuss.px4.io/ronron/32/21668_2.png) [@ronron](https://discuss.px4.io/u/ronron)\
**Post date:** [June 14, 2026, 10:16pm UTC](https://discuss.px4.io/t/multiple-arkflow-optical-flow/43083/3 "2026-06-14T22:16:39Z")

</div>

Following up on this thread, my scenario is slightly different from redundancy. I’d like to use one OF sensor facing down and one facing forward, so that when ground texture is poor the forward-facing sensor can compensate.

Is this something EKF2 could support, or does the measurement model’s dependency on `range_to_ground` make a forward-facing sensor a fundamentally bad fit for the flow fusion path? Would it make more sense to feed the forward OF as an EV\_VEL source with a paired forward rangefinder for scale?

Any pointers on what would need to change, or whether this is even worth pursuing, would be much appreciated!

---

<div class="post-metadata">

**Author:** ![Naor\_Waiss](https://discuss.px4.io/user_avatar/discuss.px4.io/naor_waiss/32/20893_2.png) [@Naor\_Waiss](https://discuss.px4.io/u/Naor_Waiss)\
**Post date:** [June 15, 2026, 2:49pm UTC](https://discuss.px4.io/t/multiple-arkflow-optical-flow/43083/4 "2026-06-15T14:49:31Z")

</div>

Hi, how did you switch between instances? I am also trying to work with two EKF2 instances: one receiving External Vision (EV) data from VIO, and a clean one (without any external sensors). However, when I try to force the EKF2 selector to switch between them, the Commander throws a lot of `nav_error` messages