gagandeepreehal/minuszero-indian-autonomous-driving-dataset

Minus Zero Indian Urban Autonomous Driving Dataset

Overview

This dataset provides original multicamera autonomous-driving recordings in MCAP format. It is designed for non-commercial research on surround-view perception, temporal and cross-camera synchronization, H.265 video pipelines, localization, GNSS/pose integration, and robotics data tooling.

Recordings include camera and GNSS/pose streams, with machine-state telemetry present in a small subset. Camera availability can change within a recording, so identities and timestamps must be validated from decoded messages.

The dataset is public for personal, educational, and research use under CC BY-NC 4.0. Commercial use is prohibited.

Key facts

Property	Value
MCAP files	1,469
Total bytes	3,196,911,932,060
Approximate download size	3.197 TB
Total indexed camera duration	85:15:58.375 (306,958.375 seconds)
Camera configuration	Multiple camera streams; recording-dependent
Container	MCAP
Camera payload	H.265/HEVC Annex-B access units
Labels/annotations	None
Authored train/validation/test split	None; the manifest viewer's technical train label is not a benchmark split

Dataset structure

The dataset payload consists of original .mcap recordings. The main branch also contains this dataset card and integrity/provenance control artifacts; it does not publish MP4 or Parquet media derivatives, extracted frames, point-cloud derivatives, or annotations. The release does not include an external camera-synchronization index.

README.md
manifest.jsonl
verification.json
data/
  <YYYY-MM-DD>/
    <recording>.mcap

Each recording contains multiple camera streams, but individual messages may contain only a subset of the recording's camera identities. Camera access units are stored inside autonome.CompressedVideos protobuf messages. An H.265 access unit inside MCAP is not an MP4 file; users who need conventional video files can extract and remux the access units without re-encoding when framing and codec parameter sets are handled correctly.

The repository tree, MCAP channels, embedded protobuf schemas, and decoded values are authoritative. Recording duration, resolution, frame cadence, available sensors, and schema details can differ between files; do not infer these properties from filenames alone.

Recording duration

The 1,469 recordings contain 85:15:58.375 of indexed camera data in total. Duration is measured for each MCAP as the autonome.CompressedVideos log-time span from its first indexed camera message to its last, then summed. It is not the wall-clock interval between separate recordings. Values below are rounded to the nearest millisecond and use hours:minutes:seconds format.

Recording date	MCAP files	Recording duration
2025-05-15	1	0:00:50.507
2025-05-19	1	0:17:27.900
2025-05-20	15	3:45:05.604
2025-05-22	6	1:36:29.909
2025-06-03	17	1:28:04.794
2025-06-05	8	0:56:03.102
2025-06-09	10	0:19:44.306
2025-06-10	25	0:11:52.594
2025-06-11	38	1:09:41.447
2025-06-12	1	1:36:09.802
2025-06-13	81	0:39:46.850
2025-06-18	1	0:02:07.589
2025-06-24	12	0:38:59.399
2025-06-25	2	0:57:57.898
2025-06-27	19	2:31:17.926
2025-06-30	8	1:48:54.591
2025-07-02	24	1:38:39.816
2025-07-03	13	0:28:46.401
2025-07-04	22	0:11:58.967
2025-07-07	49	1:06:54.169
2025-07-08	5	0:29:40.882
2025-07-10	20	1:29:30.558
2025-07-11	45	1:09:48.097
2025-07-16	11	0:09:11.765
2025-07-21	1	0:02:59.288
2025-07-22	31	0:32:33.960
2025-07-23	36	0:27:08.455
2025-07-24	10	0:06:07.136
2025-07-28	51	0:30:18.918
2025-07-30	78	0:51:35.688
2025-08-04	66	0:42:53.313
2025-08-05	24	0:09:09.358
2025-08-06	29	0:40:24.198
2025-08-07	18	0:21:15.927
2025-08-08	7	0:12:54.947
2025-08-11	17	0:06:45.552
2025-08-12	29	1:43:30.480
2025-08-13	32	0:12:21.906
2025-08-14	8	0:02:50.069
2025-08-18	11	0:04:00.832
2025-08-19	8	0:02:52.697
2025-08-20	8	0:02:54.802
2025-08-21	6	0:02:15.008
2025-08-28	31	0:13:04.573
2025-08-29	1	0:53:27.299
2025-09-01	3	1:47:59.685
2025-09-02	35	1:07:10.762
2025-09-03	3	0:37:57.735
2025-09-04	44	0:19:09.119
2025-09-05	24	0:27:43.996
2025-09-08	1	0:05:11.300
2025-09-09	70	0:34:59.478
2025-09-10	11	0:05:05.149
2025-09-11	14	0:05:02.217
2025-09-12	25	0:14:16.437
2025-09-15	25	0:44:45.926
2025-09-16	3	1:09:44.344
2025-09-17	6	4:17:13.073
2025-09-18	3	5:09:41.988
2025-09-19	52	0:50:30.905
2025-09-21	7	2:32:54.463
2025-09-23	3	2:48:09.877
2025-09-25	4	5:01:05.467
2025-09-26	4	5:50:53.184
2025-09-29	37	0:43:43.239
2025-09-30	12	0:04:24.646
2025-10-01	22	0:08:13.648
2025-10-06	12	1:30:15.195
2025-10-08	7	0:41:55.590
2025-10-10	18	0:09:06.329
2025-10-14	5	0:01:49.092
2025-10-15	4	0:01:32.207
2025-10-22	4	2:02:03.389
2025-10-23	6	1:14:51.284
2025-10-24	4	5:44:18.059
2025-10-27	5	5:09:49.229
2025-10-31	10	0:06:42.516
2025-11-03	3	0:02:06.323
2025-11-04	4	0:01:31.053
2025-11-07	3	0:01:02.895
2025-11-11	3	0:24:37.033
2025-11-12	3	0:01:11.697
2025-11-13	16	0:07:29.410
2025-11-17	18	0:23:11.155
Total	1,469	85:15:58.375

Getting started

Use selective downloads for initial exploration. The examples below keep the native MCAP representation intact.

Downloading

Install the Hugging Face client:

python -m pip install -U huggingface_hub

Download every MCAP file:

HF_REPO_ID="gagandeepreehal/minuszero-indian-autonomous-driving-dataset"
hf download "$HF_REPO_ID" \
  --repo-type dataset \
  --include "*.mcap" \
  --local-dir ./multicamera-mcap

The complete collection is approximately 3.197 TB before local filesystem overhead. Use --dry-run and narrow --include patterns to inspect and select only the dates or recordings you need before starting a transfer.

Download one recording from Python:

from huggingface_hub import hf_hub_download

mcap_path = hf_hub_download(
    repo_id="gagandeepreehal/minuszero-indian-autonomous-driving-dataset",
    filename="data/2025-05-15/2025-05-15_12-13-08.428.mcap",
    repo_type="dataset",
)
print(mcap_path)

Inspecting channels and messages

Install the MCAP readers:

python -m pip install -U mcap mcap-protobuf-support

List schemas and topics without decoding payloads:

from mcap.reader import make_reader

path = "<path/to/recording.mcap>"

with open(path, "rb") as stream:
    reader = make_reader(stream)
    seen = set()
    for schema, channel, _message in reader.iter_messages(log_time_order=False):
        key = (channel.topic, schema.name if schema else None)
        if key not in seen:
            print(key)
            seen.add(key)

Decode protobuf messages using their embedded descriptors:

from mcap.reader import make_reader
from mcap_protobuf.decoder import DecoderFactory

path = "<path/to/recording.mcap>"

with open(path, "rb") as stream:
    reader = make_reader(stream, decoder_factories=[DecoderFactory()])
    for schema, channel, message, decoded in reader.iter_decoded_messages():
        print(schema.name, channel.topic, message.log_time, type(decoded).__name__)

Before decoding or remuxing video, validate payload/name/timestamp cardinality, camera identity stability, timestamp monotonicity, access-unit framing, and codec parameter sets across the complete recording.

Multi-camera synchronization

Synchronize cameras using recorded timestamps and stable camera identities; confirm each timestamp field's clock domain and precision from decoded values and producer conventions. Do not align cameras using protobuf array position alone, decoded-frame index, filename order, or an assumed constant frame rate.

Consumers should expect and explicitly handle:

• messages that contain only a subset of the recording's cameras;
• cameras that begin late, end early, or temporarily disappear;
• dropped frames, timestamp gaps, or frame-ID discontinuities;
• recording-level camera topology derived from the union of decoded camera identities across the complete message stream;
• calibration or schema changes between recordings.

For learning pipelines, first build an external index that records source file, camera name, source timestamp, calibration identity, and any detected gaps. Such an index is not included in this direct-MCAP release.

Channels and schema reference

Channel availability varies by recording. Each topic appears once below, with file coverage and indexed message counts aggregated across the release.

Topic	Protobuf schema	Files	Indexed messages	Information available
gnss	autonome.Pose	1,469	15,317,896	Live GNSS/pose output with global/local position, velocity, orientation, uncertainty, fix status, and validity fields.
machine_state_mz-car	eCAL.pb.mma.State	6	3,261	Host resource telemetry: CPU, memory, disks, networks, temperature, processes, core count, and operating system.
rec_cameras	autonome.CompressedVideos	1,469	3,055,150	Recorded H.265/HEVC camera access units plus camera identity, timing, dimensions, and calibration fields.
rec_gnss	autonome.Pose	1,469	3,055,130	Recorded GNSS/pose path. It can share autonome.Pose with gnss, but topic identity and MCAP timestamps remain distinct.

gnss and rec_gnss are distinct live and recorded paths even though they share the autonome.Pose schema. rec_cameras carries the recorded autonome.CompressedVideos stream. Use the MCAP topic, channel metadata, timestamps, and decoded message together.

How to read the protobuf tables

The tables below show fields common across the embedded layouts for each schema. Some recordings contain additional descriptor-defined fields that are not listed here. Descriptor presence does not prove that every message populates a field. Repeated arrays can be empty and parallel arrays can differ in length, so consumers must validate cardinality per message.

Units explicitly encoded in field names are reported below. Other units, coordinate frames, clock domains, calibration-vector order, matrix layout, enum intent, and numeric ranges are producer conventions unless the descriptor states otherwise. Each MCAP's embedded descriptor and actual decoded values are authoritative.

Common protobuf schema fields

autonome.CompressedVideos — common fields (1,469 files)

Schema encoding: protobuf.

Descriptor files:

• google/protobuf/timestamp.proto — package google.protobuf, syntax proto3, dependencies: none
• autonome/CompressedVideos.proto — package autonome, syntax proto3, dependencies: google/protobuf/timestamp.proto

google.protobuf.Timestamp

Defined in google/protobuf/timestamp.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	seconds	optional	int64	Whole seconds in a protobuf timestamp.
2	nanos	optional	int32	Nanosecond fraction of a protobuf timestamp; consumers should validate its protobuf-defined range.

autonome.CompressedVideos

Defined in autonome/CompressedVideos.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	rec_frame_id	optional	int64	Recorder-assigned frame identifier.
2	rec_timestamp	optional	google.protobuf.Timestamp	Recorder timestamp.
3	timestamp	optional	google.protobuf.Timestamp	Message or sensor timestamp; its clock domain is producer-defined.
4	format	optional	string	Payload format/codec label supplied by the producer.
5	frame_id	optional	int64	Producer-assigned frame identifier.
6	data	repeated	bytes	H.265/HEVC Annex-B access-unit bytes; array position must be matched to camera identity and timestamps.
7	name	repeated	string	Camera identity strings corresponding to indexed payloads; validate cardinality and identity stability.
8	width	optional	int32	Width; exact units, range, and producer semantics are not further encoded by the descriptor.
9	height	optional	int32	Height; exact units, range, and producer semantics are not further encoded by the descriptor.
10	number_of_cameras	optional	int32	Camera count reported in this message; it need not equal recording-level topology.
11	camera_intrinsic	repeated	autonome.CompressedVideos.CameraIntrinsic	Per-camera intrinsic calibration records; align by the producer's array convention.
12	camera_extrinsic	repeated	autonome.CompressedVideos.CameraExtrinsic	Per-camera extrinsic calibration records; align by the producer's array convention.

autonome.CompressedVideos.CameraIntrinsic

Defined in autonome/CompressedVideos.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	focal_length_x	optional	float	Horizontal focal length, normally in pixels; the descriptor does not encode the calibration convention.
2	focal_length_y	optional	float	Vertical focal length, normally in pixels; the descriptor does not encode the calibration convention.
3	optical_center_x	optional	float	Horizontal principal-point coordinate, normally in pixels; convention is producer-defined.
4	optical_center_y	optional	float	Vertical principal-point coordinate, normally in pixels; convention is producer-defined.
5	rmse	optional	float	Rmse; exact units, range, and producer semantics are not further encoded by the descriptor.
6	skew	optional	float	Skew; exact units, range, and producer semantics are not further encoded by the descriptor.
7	distortion_coeffs	repeated	double	Ordered lens-distortion coefficients; model, order, and expected length are not encoded by protobuf.
8	width	optional	float	Width; exact units, range, and producer semantics are not further encoded by the descriptor.
9	height	optional	float	Height; exact units, range, and producer semantics are not further encoded by the descriptor.

autonome.CompressedVideos.CameraExtrinsic

Defined in autonome/CompressedVideos.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	rotation_vector	repeated	float	Ordered extrinsic rotation vector; parameterization, axis order, direction, and expected length are not encoded by protobuf.
2	translation_vector	repeated	float	Ordered extrinsic translation vector; units, axis order, direction, and expected length are not encoded by protobuf.

autonome.Pose — common fields (1,469 files)

Schema encoding: protobuf.

Descriptor files:

• google/protobuf/timestamp.proto — package google.protobuf, syntax proto3, dependencies: none
• Pose.proto — package autonome, syntax proto3, dependencies: google/protobuf/timestamp.proto

google.protobuf.Timestamp

Defined in google/protobuf/timestamp.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	seconds	optional	int64	Whole seconds in a protobuf timestamp.
2	nanos	optional	int32	Nanosecond fraction of a protobuf timestamp; consumers should validate its protobuf-defined range.

autonome.Pose

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	rec_frame_id	optional	int64	Recorder-assigned frame identifier.
2	rec_timestamp	optional	google.protobuf.Timestamp	Recorder timestamp.
3	timestamp	optional	google.protobuf.Timestamp	Message or sensor timestamp; its clock domain is producer-defined.
4	time_utc	optional	google.protobuf.Timestamp	UTC timestamp reported by the pose source.
5	frame_id	optional	uint64	Producer-assigned frame identifier.
6	position_type_info	optional	autonome.Pose.PositionTypeInfo	Structured or enumerated autonome.Pose.PositionTypeInfo value; see its message/enum definition below.
7	lat_lon_ht	optional	autonome.Pose.LatLonHt	Structured or enumerated autonome.Pose.LatLonHt value; see its message/enum definition below.
8	enu	optional	autonome.Pose.ENU	Structured or enumerated autonome.Pose.ENU value; see its message/enum definition below.
9	local_lat_lon_ht	optional	autonome.Pose.LocalLatLonHt	Structured or enumerated autonome.Pose.LocalLatLonHt value; see its message/enum definition below.
10	local_enu	optional	autonome.Pose.LocalENU	Structured or enumerated autonome.Pose.LocalENU value; see its message/enum definition below.
11	velocity	optional	autonome.Pose.Velocity	Structured or enumerated autonome.Pose.Velocity value; see its message/enum definition below.
12	orientation	optional	autonome.Pose.Orientation	Structured or enumerated autonome.Pose.Orientation value; see its message/enum definition below.
13	orientation_error	optional	autonome.Pose.OrientationError	Structured or enumerated autonome.Pose.OrientationError value; see its message/enum definition below.
14	position_error	optional	autonome.Pose.PositionError	Structured or enumerated autonome.Pose.PositionError value; see its message/enum definition below.
15	vehicle_ref_pt	optional	autonome.Pose.VehicleReferencePoint	Structured or enumerated autonome.Pose.VehicleReferencePoint value; see its message/enum definition below.
16	is_valid	optional	bool	Producer validity flag for the enclosing record.

Enum autonome.Pose.SpatialReferenceSystem

Number	Value
0	SRS_UNKNOWN
1	SRS_WEB_MERCATOR
2	SRS_WORLD_MERCATOR

Enum autonome.Pose.VehicleReferencePoint

Number	Value
0	REF_UNKNOWN
1	REF_ANTENNA_FRONT
2	REF_ANTENNA_REAR
3	REF_IMU
4	REF_WHEEL_AXLE_FRONT
5	REF_WHEEL_AXLE_REAR
6	REF_CAMERA_FRONT
7	REF_CAMERA_REAR
8	REF_LIDAR

autonome.Pose.PositionTypeInfo

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	rtk_condition	optional	uint32	Producer-defined numeric RTK condition/status.
2	correction_age_s	optional	float	Correction age s. Value in seconds.
3	position_fix_type	optional	uint32	Producer-defined numeric GNSS position-fix classification.

autonome.Pose.LatLonHt

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	latitude_deg	optional	double	Latitude deg. Angle or geographic coordinate in degrees.
2	longitude_deg	optional	double	Longitude deg. Angle or geographic coordinate in degrees.
3	height_m	optional	double	Height m. Value in metres.

autonome.Pose.ENU

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	east_m	optional	double	East m. Value in metres.
2	north_m	optional	double	North m. Value in metres.
3	up_m	optional	double	Up m. Value in metres.
4	ref_system_type	optional	autonome.Pose.SpatialReferenceSystem	Spatial reference-system enum for ENU coordinates.

autonome.Pose.LocalLatLonHt

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	local_datum_id	optional	string	Identifier for the local positioning datum.
2	latitude_deg	optional	double	Latitude deg. Angle or geographic coordinate in degrees.
3	longitude_deg	optional	double	Longitude deg. Angle or geographic coordinate in degrees.
4	height_m	optional	double	Height m. Value in metres.

autonome.Pose.LocalENU

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	local_east_m	optional	double	Local east m. Value in metres.
2	local_north_m	optional	double	Local north m. Value in metres.
3	local_datum_height_m	optional	double	Local datum height m. Value in metres.

autonome.Pose.Velocity

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	east_mps	optional	double	East mps. Value in metres per second.
2	north_mps	optional	double	North mps. Value in metres per second.
3	up_mps	optional	double	Up mps. Value in metres per second.

autonome.Pose.Orientation

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	roll_rad	optional	double	Roll rad. Angle in radians.
2	pitch_rad	optional	double	Pitch rad. Angle in radians.
3	yaw_rad	optional	double	Yaw rad. Angle in radians.

autonome.Pose.OrientationError

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	pitch_variance_rad2	optional	double	Pitch variance rad2. Variance in radians squared.
2	yaw_variance_rad2	optional	double	Yaw variance rad2. Variance in radians squared.
3	roll_variance_rad2	optional	double	Roll variance rad2. Variance in radians squared.

autonome.Pose.PositionError

Defined in Pose.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	position_rms	optional	double	Position root-mean-square error; units are not encoded in the field name or descriptor.
2	east_sigma_m	optional	double	East sigma m. Value in metres.
3	north_sigma_m	optional	double	North sigma m. Value in metres.
4	up_sigma_m	optional	double	Up sigma m. Value in metres.
5	covariance_en	optional	double	East/north covariance; unit convention is producer-defined.
6	pdop	optional	float	Position dilution of precision.
7	hdop	optional	float	Horizontal dilution of precision.
8	vdop	optional	float	Vertical dilution of precision.

eCAL.pb.mma.State (6 files)

Descriptor size: 939 bytes. Schema encoding: protobuf.

Descriptor files:

• ecal/app/pb/mma/mma.proto — package eCAL.pb.mma, syntax proto3, dependencies: none

eCAL.pb.mma.Disk

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	name	optional	string	Name; exact units, range, and producer semantics are not further encoded by the descriptor.
2	read	optional	double	Read; exact units, range, and producer semantics are not further encoded by the descriptor.
3	write	optional	double	Write; exact units, range, and producer semantics are not further encoded by the descriptor.
4	available	optional	uint64	Available; exact units, range, and producer semantics are not further encoded by the descriptor.
5	capacity	optional	uint64	Capacity; exact units, range, and producer semantics are not further encoded by the descriptor.

eCAL.pb.mma.Network

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	name	optional	string	Name; exact units, range, and producer semantics are not further encoded by the descriptor.
2	send	optional	double	Send; exact units, range, and producer semantics are not further encoded by the descriptor.
3	receive	optional	double	Receive; exact units, range, and producer semantics are not further encoded by the descriptor.
4	ip_address	optional	string	Ip address; exact units, range, and producer semantics are not further encoded by the descriptor.

eCAL.pb.mma.Memory

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	total	optional	uint64	Total; exact units, range, and producer semantics are not further encoded by the descriptor.
2	available	optional	uint64	Available; exact units, range, and producer semantics are not further encoded by the descriptor.

eCAL.pb.mma.Temperature

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	cpu	optional	float	Cpu; exact units, range, and producer semantics are not further encoded by the descriptor.

eCAL.pb.mma.ProcessesMemory

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	current_working_set_size	optional	uint64	Current working set size; exact units, range, and producer semantics are not further encoded by the descriptor.
2	peak_working_set_size	optional	uint64	Peak working set size; exact units, range, and producer semantics are not further encoded by the descriptor.

eCAL.pb.mma.ProcessesCPU

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	cpu_kernel_time	optional	uint64	Cpu kernel time; exact units, range, and producer semantics are not further encoded by the descriptor.
2	cpu_user_time	optional	uint64	Cpu user time; exact units, range, and producer semantics are not further encoded by the descriptor.
3	cpu_creation_time	optional	uint64	Cpu creation time; exact units, range, and producer semantics are not further encoded by the descriptor.
4	cpu_load	optional	float	Cpu load; exact units, range, and producer semantics are not further encoded by the descriptor.

eCAL.pb.mma.Process

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	name	optional	string	Name; exact units, range, and producer semantics are not further encoded by the descriptor.
2	id	optional	uint32	Id; exact units, range, and producer semantics are not further encoded by the descriptor.
3	commandline	optional	string	Commandline; exact units, range, and producer semantics are not further encoded by the descriptor.
4	memory	optional	eCAL.pb.mma.ProcessesMemory	Structured or enumerated eCAL.pb.mma.ProcessesMemory value; see its message/enum definition below.
5	cpu	optional	eCAL.pb.mma.ProcessesCPU	Structured or enumerated eCAL.pb.mma.ProcessesCPU value; see its message/enum definition below.
6	debug_flag	optional	bool	Boolean debug flag flag/state; exact producer semantics are not further encoded.
7	user	optional	string	User; exact units, range, and producer semantics are not further encoded by the descriptor.

eCAL.pb.mma.State

Defined in ecal/app/pb/mma/mma.proto. Map-entry: false. Oneofs: none.

No.	Field	Cardinality	Protobuf type	Information available
1	cpu_load	optional	double	Cpu load; exact units, range, and producer semantics are not further encoded by the descriptor.
2	memory	optional	eCAL.pb.mma.Memory	Structured or enumerated eCAL.pb.mma.Memory value; see its message/enum definition below.
3	disks	repeated	eCAL.pb.mma.Disk	Structured or enumerated eCAL.pb.mma.Disk value; see its message/enum definition below.
4	networks	repeated	eCAL.pb.mma.Network	Structured or enumerated eCAL.pb.mma.Network value; see its message/enum definition below.
5	temperature	optional	eCAL.pb.mma.Temperature	Structured or enumerated eCAL.pb.mma.Temperature value; see its message/enum definition below.
6	process	repeated	eCAL.pb.mma.Process	Structured or enumerated eCAL.pb.mma.Process value; see its message/enum definition below.
7	number_of_cpu_cores	optional	uint32	Number of cpu cores; exact units, range, and producer semantics are not further encoded by the descriptor.
8	operating_system	optional	string	Operating system; exact units, range, and producer semantics are not further encoded by the descriptor.

Dataset Viewer

MCAP is a robotics log container and the embedded camera video is not natively previewable on the Hub. The configured manifest.jsonl appears in Data Studio as a 1,469-row technical train split containing repository paths, SHA-256 hashes, sizes, and source-provenance fields. It does not decode MCAP messages or provide browser video playback. Download the MCAP files and use MCAP-aware tools locally.

Intended uses

Appropriate non-commercial uses include:

• multi-camera perception, synchronization, and sensor-fusion research;
• autonomous-driving video understanding and robotics experiments;
• MCAP reader, indexing, or data-pipeline development;
• H.265 extraction, timestamp analysis, and remuxing experiments;
• non-commercial teaching, demonstrations, and personal projects.

The dataset is not validated for production driving systems, operational safety decisions, surveillance, identity recognition, or evaluation claims beyond the documented contents.

Limitations and known risks

• The dataset is unannotated and does not provide benchmark ground truth.
• Camera topology must be derived from complete decoded message streams; one message may expose only a subset of cameras.
• A camera's absence from one message does not prove absence from the recording.
• Geographic, environmental, temporal, traffic, and hardware coverage may be uneven.
• Frames can be missing, delayed, duplicated, corrupted, or irregularly spaced.
• Camera calibration and protobuf schema details can differ between recordings.
• H.265 decoding support varies by platform and software build.
• Channel presence does not guarantee that every field is populated; payloads and file integrity require per-recording validation.
• Source timestamps should be preserved; synthetic constant-frame-rate conversion can lose timing and synchronization information.

Privacy and responsible use

Recordings may have been captured in public-road environments and may incidentally contain people, faces, vehicle registration plates, residences, businesses, road signs, or precise GNSS/location traces. Do not assume that the data is anonymized.

Users are responsible for reviewing applicable privacy, data-protection, publicity, export-control, and local recording laws before downloading, redistributing, publishing examples, or training models. Avoid re-identification, tracking individuals, exposing sensitive locations, or combining this dataset with external data in ways that increase privacy risk.

If you believe a recording contains sensitive, unlawful, or improperly included material, open a repository discussion with the repository path and a concise description.

License

This dataset is released under the Creative Commons Attribution-NonCommercial 4.0 International license (CC BY-NC 4.0).

• Attribution is required.
• Commercial use is prohibited.
• Sharing and adaptation are permitted for non-commercial purposes under the license terms.
• Indicate whether you modified the material and include a link to the license.

The intended audience is personal, educational, and research users. The legal permissions and conditions are defined by CC BY-NC 4.0; this card does not replace the license text. No warranties are provided, and other rights such as privacy, publicity, or trademark rights may still apply.

Citation

Cite this dataset as:

@dataset{minuszero_indian_autonomous_driving_2026,
  author    = {Gagandeep Reehal},
  title     = {Minus Zero Indian Urban Autonomous Driving Dataset},
  year      = {2026},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/gagandeepreehal/minuszero-indian-autonomous-driving-dataset}
}

Maintenance

Dataset questions, correction requests, and takedown reports: repository discussions.

Consumers should pin an immutable commit revision when reproducibility matters.