fctl provides a powerful interface to the Formant data backend. You can use the query command to get data from your robots at any point in time. Similar to tail you can also follow in real-time as data flows into Formant.
Prerequisite
First, make sure you've installed fctl
Query data
The first thing you may want to do is query data. One important note, this command queries data from all your devices so use with care, however it is extremely powerful:
fctl query
You'll see table style output with all the device streams from the last 10 minutes. Please note that the timestamp returned by fctl is in UTC.
TIMESTAMP (UTC) | DEVICE NAME | STREAM NAME | TYPE | DATA | TAGS
+-------------------------------+-------------+--------------------------------------------+---------+------------------------------------------------------------------+------+
2020-06-03T22:04:45.928-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:05:01.137-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:05:16.294-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:05:31.543-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:05:46.689-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:06:01.84-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:06:16.987-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:06:32.137-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:06:47.286-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:07:02.518-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:07:17.668-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:07:32.918-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:07:48.162-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:08:03.324-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:08:18.494-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:08:33.641-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:08:48.792-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:09:03.944-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:09:19.219-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:09:34.372-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:09:49.513-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:10:04.671-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:10:19.822-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:10:34.981-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:10:50.133-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:11:05.277-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:11:20.578-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:11:35.736-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:11:50.881-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:12:06.034-07:00 | pi.b.001 | $.agent.asset_disk_buffer_fill_percent | numeric | 0 | {}
2020-06-03T22:13:57.891-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:14:07.905-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:14:17.924-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:14:27.938-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:14:37.952-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:04:39-07:00 | nano.001 | test.numeric | numeric | 15281 | {}
2020-06-03T22:04:40-07:00 | nano.001 | test.numeric | numeric | 1685 | {}
2020-06-03T22:04:41-07:00 | nano.001 | test.numeric | numeric | 29143 | {}
2020-06-03T22:04:42-07:00 | nano.001 | test.numeric | numeric | 7136 | {}
2020-06-03T22:04:43-07:00 | nano.001 | test.numeric | numeric | 11967 | {}
2020-06-03T22:04:44-07:00 | nano.001 | test.numeric | numeric | 12119 | {}
2020-06-03T22:04:45-07:00 | nano.001 | test.numeric | numeric | 15907 | {}
2020-06-03T22:04:46-07:00 | nano.001 | test.numeric | numeric | 4776 | {}
2020-06-03T22:04:47-07:00 | nano.001 | test.numeric | numeric | 9588 | {}
2020-06-03T22:04:48-07:00 | nano.001 | test.numeric | numeric | 3825 | {}
2020-06-03T22:04:49-07:00 | nano.001 | test.numeric | numeric | 13227 | {}
2020-06-03T22:04:50-07:00 | nano.001 | test.numeric | numeric | 22390 | {}
2020-06-03T22:04:51-07:00 | nano.001 | test.numeric | numeric | 18780 | {}
2020-06-03T22:04:52-07:00 | nano.001 | test.numeric | numeric | 29685 | {}
2020-06-03T22:04:53-07:00 | nano.001 | test.numeric | numeric | 31427 | {}
2020-06-03T22:04:54-07:00 | nano.001 | test.numeric | numeric | 25891 | {}
2020-06-03T22:04:55-07:00 | nano.001 | test.numeric | numeric | 5627 | {}
2020-06-03T22:04:56-07:00 | nano.001 | test.numeric | numeric | 14974 | {}
2020-06-03T22:04:57-07:00 | nano.001 | test.numeric | numeric | 23917 | {}
2020-06-03T22:04:58-07:00 | nano.001 | test.numeric | numeric | 19416 | {}
2020-06-03T22:04:59-07:00 | nano.001 | test.numeric | numeric | 7700 | {}
2020-06-03T22:05:00-07:00 | nano.001 | test.numeric | numeric | 3994 | {}
2020-06-03T22:05:01-07:00 | nano.001 | test.numeric | numeric | 4316 | {}
2020-06-03T22:05:02-07:00 | nano.001 | test.numeric | numeric | 10268 | {}
2020-06-03T22:05:03-07:00 | nano.001 | test.numeric | numeric | 2069 | {}
2020-06-03T22:05:04-07:00 | nano.001 | test.numeric | numeric | 16902 | {}
2020-06-03T22:05:06-07:00 | nano.001 | test.numeric | numeric | 24766 | {}
2020-06-03T22:05:07-07:00 | nano.001 | test.numeric | numeric | 12865 | {}
2020-06-03T22:05:08-07:00 | nano.001 | test.numeric | numeric | 891 | {}
2020-06-03T22:05:09-07:00 | nano.001 | test.numeric | numeric | 12918 | {}
2020-06-03T22:05:10-07:00 | nano.001 | test.numeric | numeric | 14322 | {}
Query just one stream
You can also query by stream name across all your devices.
fctl query --stream $.agent.health
TIMESTAMP (UTC) | DEVICE NAME | STREAM NAME | TYPE | DATA | TAGS
+-------------------------------+-------------+----------------+--------+--------------------------+------+
2020-06-03T22:12:28.816-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:12:29.817-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:12:30.818-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:12:31.819-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:12:32.822-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:12:33.823-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:12:34.827-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:12:35.828-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:18.873-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:19.874-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:20.874-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:21.874-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:22.874-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:23.875-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:24.875-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:25.875-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:26.875-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:22:27.876-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
You can see how you can query data across devices for the same stream. The above query returns data from the last 10 minutes.
Query for a device
You can also query data for a single device
fctl query --device nano.001
TIMESTAMP (UTC) | DEVICE NAME | STREAM NAME | TYPE | DATA | TAGS
+-------------------------------+-------------+--------------------------------------------+---------+------------------------------------------------------------------+------+
2020-06-03T22:24:08.771-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:24:18.785-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:24:28.8-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:24:38.814-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:24:48.828-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:24:58.843-07:00 | nano.001 | $.ros.node_online | bitset | {"keys":["/formant_bridge_node","/rosout"],"values":[true,true]} | {}
2020-06-03T22:15:07-07:00 | nano.001 | test.numeric | numeric | 27671 | {}
2020-06-03T22:15:08-07:00 | nano.001 | test.numeric | numeric | 2311 | {}
2020-06-03T22:15:09-07:00 | nano.001 | test.numeric | numeric | 21811 | {}
2020-06-03T22:15:10-07:00 | nano.001 | test.numeric | numeric | 17388 | {}
2020-06-03T22:15:12-07:00 | nano.001 | test.numeric | numeric | 6457 | {}
2020-06-03T22:15:13-07:00 | nano.001 | test.numeric | numeric | 16089 |
All the devices streams will appear here.
Query with start and end time
Sometimes you will want to query historical data or a small slice. You can do that by using UTC RFC3339 timestamps:
fctl query --start 2020-06-04T05:15:13 --end 2020-06-04T05:15:14
TIMESTAMP (UTC) | DEVICE NAME | STREAM NAME | TYPE | DATA | TAGS
+-------------------------------+-------------+-------------------------+---------+--------------------------+------+
2020-06-03T22:15:13.193-07:00 | pi.b.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:15:13.242-07:00 | pi.b.001 | $.host.cpu_util_percent | numeric | 21.000000000003638 | {}
2020-06-03T22:15:13.808-07:00 | pi.b.001 | $.host.load_avg_1min | numeric | 0.43 | {}
2020-06-03T22:15:13.208-07:00 | pi.b.001 | $.host.mem_used_percent | numeric | 41.94699378522908 | {}
2020-06-03T22:15:13.6-07:00 | pi.b.001 | $.host.net_util_rx_mbps | numeric | 0.007808 | {}
2020-06-03T22:15:13.599-07:00 | pi.b.001 | $.host.net_util_tx_mbps | numeric | 0.026512 | {}
2020-06-03T22:15:13.76-07:00 | nano.001 | $.agent.health | health | {"status":"operational"} | {}
2020-06-03T22:15:13.526-07:00 | nano.001 | $.host.cpu_util_percent | numeric | 2.250000007916242 | {}
2020-06-03T22:15:13.075-07:00 | nano.001 | $.host.load_avg_1min | numeric | 0.02 | {}
2020-06-03T22:15:13.471-07:00 | nano.001 | $.host.mem_used_percent | numeric | 68.11337632947738 | {}
2020-06-03T22:15:13.031-07:00 | nano.001 | $.host.net_util_rx_mbps | numeric | 0.004112 | {}
2020-06-03T22:15:13.031-07:00 | nano.001 | $.host.net_util_tx_mbps | numeric | 0.012072 | {}
2020-06-03T22:15:13-07:00 | nano.001 | test.numeric | numeric | 16089 | {}
Query by tags
You can also query by tags. We have updated one of our streams to have the value: actual tag:
fctl query --tags 'value:actual'
TIMESTAMP (UTC) | DEVICE NAME | STREAM NAME | TYPE | DATA | TAGS
+---------------------------+-------------+-------------+---------+------------+--------------------+
2020-06-03T22:32:48-07:00 | nano.001 | speed | numeric | 0.459796 | {"value":"actual"}
2020-06-03T22:32:50-07:00 | nano.001 | speed | numeric | 0.722872 | {"value":"actual"}
2020-06-03T22:32:51-07:00 | nano.001 | speed | numeric | 0.854409 | {"value":"actual"}
2020-06-03T22:32:52-07:00 | nano.001 | speed | numeric | 0.985947 | {"value":"actual"}
2020-06-03T22:32:53-07:00 | nano.001 | speed | numeric | 0.117485 | {"value":"actual"}
2020-06-03T22:32:54-07:00 | nano.001 | speed | numeric | 0.249023 | {"value":"actual"}
2020-06-03T22:32:55-07:00 | nano.001 | speed | numeric | 0.380561 | {"value":"actual"}
2020-06-03T22:32:56-07:00 | nano.001 | speed | numeric | 0.512098 | {"value":"actual"}
2020-06-03T22:32:57-07:00 | nano.001 | speed | numeric | 0.643636 | {"value":"actual"}
2020-06-03T22:32:58-07:00 | nano.001 | speed | numeric | 0.775174 | {"value":"actual"}
You can also query for multiple tags
fctl query --tags 'value:actual,limit'
TIMESTAMP (UTC) | DEVICE NAME | STREAM NAME | TYPE | DATA | TAGS
+---------------------------+-------------+-------------+---------+------------+--------------------+
2020-06-03T22:35:48-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:35:49-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:35:50-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:35:51-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:35:52-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:35:53-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:35:54-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:35:55-07:00 | nano.001 | speed | numeric | 1.2 | {"value":"limit"}
2020-06-03T22:32:48-07:00 | nano.001 | speed | numeric | 0.459796 | {"value":"actual"}
2020-06-03T22:32:50-07:00 | nano.001 | speed | numeric | 0.722872 | {"value":"actual"}
2020-06-03T22:32:51-07:00 | nano.001 | speed | numeric | 0.854409 | {"value":"actual"}
2020-06-03T22:32:52-07:00 | nano.001 | speed | numeric | 0.985947 | {"value":"actual"}
2020-06-03T22:32:53-07:00 | nano.001 | speed | numeric | 0.117485 | {"value":"actual"}
2020-06-03T22:32:54-07:00 | nano.001 | speed | numeric | 0.249023 | {"value":"actual"}
2020-06-03T22:32:55-07:00 | nano.001 | speed | numeric | 0.380561 | {"value":"actual"}
2020-06-03T22:32:56-07:00 | nano.001 | speed | numeric | 0.512098 | {"value":"actual"}
Download uploaded files
fctl also allows you to download any files uploaded to image, pointcloud and file streams. For example:
fctl query --device nano.001 --stream bags --download
You'll see this output as files are downloaded.
Downloaded: 1591720348.bag
Downloaded: 1591720408.bag
Downloaded: 1591720378.bag
Downloaded: 1591720368.bag
Downloaded: 1591720358.bag
Downloaded: 1591720388.bag
Downloaded: 1591720398.bag
The files are downloaded to a directory structure of <device_name>/<stream_name>/<files>:
ls nano.001/bags/
1591720274.bag 1591720348.bag 1591720358.bag 1591720368.bag 1591720378.bag 1591720388.bag 1591720398.bag 1591720408.bag
Advanced usage with jq
jq is a powerful command line utility for JSON data. One of the featues of fctl is the ability to output data in JSON. For example:
fctl query --device nano.001 --stream test.numeric --start 2020-06-04T05:15:13 --end 2020-06-04T05:15:14 --output json
{"deviceId":"ff1e0d3b-5044-4db1-a459-c403cc2159a3","deviceName":"nano.001","timestamp":"2020-06-03T22:15:13-07:00","streamName":"test.numeric","value":"16089","type":"numeric","tags":{}}
This output is JSON, but we can prettify it with jq:
fctl query --device nano.001 --stream test.numeric --start 2020-06-04T05:15:13 --end 2020-06-04T05:15:14 --output json | jq
{
"deviceId": "ff1e0d3b-5044-4db1-a459-c403cc2159a3",
"deviceName": "nano.001",
"timestamp": "2020-06-03T22:15:13-07:00",
"streamName": "test.numeric",
"value": "16089",
"type": "numeric",
"tags": {}
}
Way better. We can also extract single fields with jq:
fctl query --device nano.001 --stream test.numeric --start 2020-06-04T05:15:13 --end 2020-06-04T05:15:14 --output json | jq '.value'
"16089"
Let's actually look at this value in real-time. fctl has an additional flag for following data -f or --follow. Let's also switch over to our speed stream:
fctl query --device nano.001 --stream speed --follow --output json | jq '.value'
"0.553797"
"0.685334"
"0.816872"
"0.94841"
"0.0799478"
"0.343023"
"1.2"
"1.2"
"1.2"
"1.2"
.
.
.
One thing we notice here immediately is that we have both the limit and actual intermixed. Let's use jq to clear that up a bit. Notice in this example we construct an output JSON with the tag value as the key:
fctl query --device nano.001 --stream speed --follow --output json | jq '{ (.tags.value): .value }'
{
"actual": "0.5319"
}
{
"actual": "0.663438"
}
{
"actual": "0.794976"
}
{
"actual": "0.926513"
}
{
"limit": "1.2"
}
{
"limit": "1.2"
}
{
"limit": "1.2"
}
{
"limit": "1.2"
}
{
"actual": "0.0580511"
}
{
"actual": "0.189589"
}
{
"actual": "0.321127"
}
{
"actual": "0.452664"
}
We can get fancier when we start using jq programs like those found in the jq directory. One note on these programs is they require the use of the --slurp flag so you won't be able to run these wit the fctl --follow flag. However, you could programatically loop using standard bash utilites.
First let's calculate the average, variance, and standard deviation of our actual speed (assume these are run from the docs directory) using the math.jq program:
fctl query --device nano.001 --start 2020-06-05T05:40:00 --end 2020-06-05T05:55:00 --stream speed --tags 'value:actual' --output json | jq -s -f jq/math.jq
{
"avg": 0.5017421673093813,
"stdev": 0.2898920770688194,
"variance": 0.08403741634727435
}
Next, let's use the filter.jq program to filter our data for any actual speed greater than 0.8:
fctl query --device nano.001 --start 2020-06-05T05:40:00 --end 2020-06-05T05:41:00 --stream speed --tags 'value:actual' --output json | jq -s -f jq/filter.jq
[
{
"deviceId": "8aced12e-4514-4e13-96c9-e881fd3e5f24",
"deviceName": "nano.001",
"timestamp": "2020-06-04T22:41:13-07:00",
"streamName": "speed",
"value": "0.751275",
"type": "numeric",
"tags": {
"hw": "jetson_nano",
"value": "actual"
}
},
{
"deviceId": "8aced12e-4514-4e13-96c9-e881fd3e5f24",
"deviceName": "nano.001",
"timestamp": "2020-06-04T22:41:14-07:00",
"streamName": "speed",
"value": "0.882812",
"type": "numeric",
"tags": {
"hw": "jetson_nano",
"value": "actual"
}
}
]
We can show how to group averages across a set of tags. In this example we'll calculate averages for all values in the speed stream grouped by the value tag. This is defined in the bucket_tags.jq
fctl query --device nano.001 --start 2020-06-05T05:40:00 --end 2020-06-05T05:42:00 --stream speed --output json | jq -s -f jq/bucket_tags.jq
[
{
"value": "actual",
"avg": 0.5111393844444444
},
{
"value": "limit",
"avg": 1.200000000000001
}
]
Similar to bucketing by tags we can also bucket by device using the bucket_device.jq program. Let's see how our averages compare across our devices:
fctl query --start 2020-06-05T06:30:00 --end 2020-06-05T06:35:00 --stream speed --tags 'value:actual' --output json | jq -s -f jq/bucket_device.jq
[
{
"value": "pi.b.001",
"avg": 0.5680556666666668
},
{
"value": "nano.001",
"avg": 0.8522845626555992
}
]
As you can see we're fans of both raspberry pi's and Nvidia jetson boards. Let's see our speed averages grouped by the tag hw with the bucet_hw.jq program:
$ fctl query --start 2020-06-05T06:40:00 --end 2020-06-05T06:45:00 --stream speed --tags 'value:actual' --output json | jq -s -f jq/bucket_hw.jq
[
{
"value": "pi_b",
"avg": 0.5098477718562874
},
{
"value": "jetson_nano",
"avg": 0.5094469515093457
}
]