Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 10 Current »

Status

COMPLETE

Impact

HIGH

Driver

Anthony Luo

Approver

Contributors

Anthony Luo

Informed

Mechanical: Conall Kingshott Nathan Green Alison Thompson
Electrical: Michael Botros Nolan Haines Mena Azab
Autonomy: Mihir Gupta Amy Hu
EFS: Aaditya Chaudhary Derek Tang Aidan Bowers (Deactivated)

Due date

Resources

References

2023-09-24 Conops Brief #1

\uD83D\uDCDA Relevant data

This follows from the discussion during the 2023-09-24 Conops Brief for the 2024 AEAC Student UAS competition brief.

\uD83D\uDCD8 Background

Our end-goal is to provide safety for the drone and it’s payload when landing in a LZ-C, which is a BVLOS landing zone with overhangs and obstacles.

Existing support

Ardupilot has native support for both object detection, as well as multiple lidar sensors. Using information synthesized from up to 10 sectors (8 around, 1 top, 1 bottom), and known INS parameters, the vehicle is able to stop at a predetermined distance from the fence regardless of the speed of the vehicle. This service is provided in all guided flight modes, as well as altitude hold.

When data is not received from all 8 sectors, empty sectors are filled in with distance from an adjacent sector (if available).

In regular operation, we only anticipate approaching obstacles facing forwards, and so sideways and rear-facing object detection is not as important.

Multiple kinds of rangefinders are supported within ardupilot, which can be found within copter documentation: Rangefinders (landing page) — Copter documentation (ardupilot.org).

\uD83C\uDF08 360 vs Unidirectional

360

Unidirectional

Description

Using a 360 degree lidar, it is possible to obtain distance and object avoidance on all sides.

Using a series of omnidirectional rangefinders, it is possible to maintain object avoidance on as many sides as sensors are used

Pros and cons

(plus) 360 object avoidance. Fairly simple configuration and setup

(minus) Cost (~ 500 CAD)

(minus) Weight (~ 200 grams)

(plus) Cost (~ 100 CAD)

(plus) Weight ( ~ 50 grams)

(minus) Harnessing needs CAN interface boards. (~ 100 CAD, 50 grams)

(minus) Potentially not all sides of the drone

(minus) May not be as comprehensive coverage as 360 lidar.

🚨 Omnidirectional Sensor Options

Buy link / datasheet

Lightware SF40/C

Lightware SF45/B

RPLidar A2

TerraRange Tower/Tower Evo

Cygbot D1

Product Page

SF40/C – Lightware Lidar Explore

SF45/B – Lightware Lidar Explore

RPlidar A2 - SLAMTEC Global Network Lidar Sensor RPlidar A2 cheap lidar scanner

TeraRanger Tower Evo | Solid state LiDAR system | Anti-collision (terabee.com)

Cygbot CygLiDAR D1 - SEN-18580 - SparkFun Electronics

Coverage (degs)

360

350/320

360

360

120

Cost (CAD)

800

450

220

1000

180

Weight (grams)

256

59

190

135

28g

Max range (m)

100

50

16

60/8

8

Min range (m)

0.2

0.2

0.2

0.5/0.75

0.2

VSupply

12v, 5v

5v

5v

12-24v

5v

Interfaces

(not all may be supported by Ardupilot)

Serial UART

Serial UART, I2C

Serial UART

Serial UART

Serial UART

OK in direct sunlight? (Per manu recc:)

Y

Y

N

Y

N

Weatherproof?

No

No

No

?

No

❗ Unidirectional Sensor Options

Since the list is incredibly exhaustive, I have pre-selected sensors which meet the minimum range (~ 8m) and cost requirements (< 100 per part).

Sensors which have not been included are:

  • Ainstein US-D1: Cost (600 each)

  • Attollo Wasp-200: Cost (530 each)

  • TF-Luna: Range (outdoors range of 3m only)

  • GY-US42v2: Range (indoor only and only 4m)

  • Hondex HE-8S: Use case (this is an underwater sonar)

  • HC-SR04: Range (2m indoor only)

  • JSN-SR04T: Consistency/reliability ( poor )

  • Leddar One Lidar: Sourcing (impossible to find)

  • Leddar Vu8: Sourcing (impossible to find) & price (1100)

  • Lightware SF1/11/20/02 (price, sourcing)

  • Maxbotix Sonar: Range (7m), use (indoor)

  • ST VL53L0X/VL53L1X: Range (2/4m)

It is worth noting that all unidirectional sensors will need to use a periph can node, which is around 4 grams per board and costs ~20 CAD.

\uD83C\uDF1F Recommendations

Pending review, the recommended solution is to use 3 Benewake TF-Mini-S series sensors or 1 Lightware SF45-B Sensor. Should the need present itself, the SF45-B may be augmented by downwards and upwards facing rangefinders. The following considerations should be reviewed:

  1. Weight

3*5 = 15g. Even considering the weight (~4g per) of the CAN interfacing boards, the weight of the benewake sensors comes out to around the same weight as the omnidirectional lidar (27 vs 59 grams), not including the non-negligible wire weight that the unidirectional sensors will have to maintain.

Both options presented here are significantly lighter than any alternatives.

  1. Packaging/form factor

Both options presented here are slimmer/sleeker than the others. Unidirectional sensors may be similar to each other, but the SF45 is significantly sleeker and easier to integrate than other sensors.

  1. Cost

The unidirectional system would total to around 180, while the omnidirectional system would be a flat price of around 450 cad. Both costs are still significantly lower than other alternatives, with the possibility of using superior sensors unidirectionally, or more sensors unidirectionally, to compensate for the relatively worse performance of the TF-Mini-S.

✅ Outcome

For 360

For Uni

Anni

  • Costs less, weighs less.

Daniel Puratich

  • Imo easier to mount a few unidirectionals then a single omnidirection because it’s easier to not obscure the sensor with other parts of the drone

  • Decision analysis is comprehensive and I agree with recommendation.

  • Cost aspect is really important!

Megan

  • agreed

  • 360 lidar would be probably in a crumple zone, at least with unidirection they can be protected better

Alison

  • only plus here is we’d only need one mount as opposed to a few, don’t think this is the biggest deal but worth mentioning

Alison

  • also agreed

  • less weight = biggest factor imo

  • cost less = also good

🗒️ Changelog

Version Date Comment
Current Version (v. 10) 2023-10-04 01:50 Mihir Gupta
v. 10 2023-10-04 01:50 Mihir Gupta
v. 9 2023-10-04 01:25 Alison Thompson
v. 8 2023-10-04 01:23 Alison Thompson
v. 7 2023-10-03 02:55 Megan Spee
attempt 2 at updating the changelog
v. 6 2023-10-03 02:53 Megan Spee
version comment
v. 5 2023-10-03 02:52 Megan Spee
v. 4 2023-10-01 15:29 Daniel Puratich
v. 3 2023-10-01 02:24 Anthony Luo
v. 2 2023-10-01 02:21 Anthony Luo
v. 1 2023-10-01 02:19 Anthony Luo
Initial version with preliminary information. No decision yet
  • No labels